data_science_final_project.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"Data_Science_final_project.ipynb","provenance":[],"collapsed_sections":[],"toc_visible":true,"mount_file_id":"1cdkZziUceBDPvPaTVVSqkCK26CmnikLb","authorship_tag":"ABX9TyP7JXnB7iptSDjYv4SH6ajA"},"kernelspec":{"name":"python3","display_name":"Python 3"},"accelerator":"GPU"},"cells":[{"cell_type":"markdown","metadata":{"id":"CYTMtIvHAc1i","colab_type":"text"},"source":["# Data Science for Cyber Security -- Final Project"]},{"cell_type":"markdown","metadata":{"id":"FSHkJfhMFnrg","colab_type":"text"},"source":["# Malware Detection (Static Analysis)"]},{"cell_type":"code","metadata":{"id":"FpNP-DghaQQo","colab_type":"code","outputId":"a0ef65bc-6d42-4493-b84e-8e01d3c6cf22","executionInfo":{"status":"ok","timestamp":1592059550532,"user_tz":-480,"elapsed":1289,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":71}},"source":["import pandas as pd\n","import numpy as np\n","import matplotlib.pyplot as plt\n","import seaborn as sns"],"execution_count":1,"outputs":[{"output_type":"stream","text":["/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n","  import pandas.util.testing as tm\n"],"name":"stderr"}]},{"cell_type":"code","metadata":{"id":"wC5IYl5a__W1","colab_type":"code","colab":{}},"source":["import tensorflow as tf\n","from tensorflow import keras\n","import glob\n","import os"],"execution_count":0,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"sFhRKlHlAXFK","colab_type":"text"},"source":["## Part1. Data Preprocessing"]},{"cell_type":"markdown","metadata":{"id":"v6ULBGR9GaeS","colab_type":"text"},"source":["### Read Data"]},{"cell_type":"code","metadata":{"id":"11lxvchIARG3","colab_type":"code","outputId":"e4700172-3481-47bc-849c-6fae1b7df36d","executionInfo":{"status":"ok","timestamp":1592059554574,"user_tz":-480,"elapsed":5313,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":204}},"source":["# .text(code) Section Header Data\n","header_df = pd.read_csv(\"https://drive.google.com/uc?id=1G69_tLvo9QluAh7LmaOA-KZWbKusXAd4\")\n","header_df.head()"],"execution_count":3,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>hash</th>\n","      <th>size_of_data</th>\n","      <th>virtual_address</th>\n","      <th>entropy</th>\n","      <th>virtual_size</th>\n","      <th>malware</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>071e8c3f8922e186e57548cd4c703a5d</td>\n","      <td>443392</td>\n","      <td>4096</td>\n","      <td>6.529624</td>\n","      <td>442984</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>33f8e6d08a6aae939f25a8e0d63dd523</td>\n","      <td>331264</td>\n","      <td>4096</td>\n","      <td>6.604314</td>\n","      <td>330784</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>b68abd064e975e1c6d5f25e748663076</td>\n","      <td>74240</td>\n","      <td>4096</td>\n","      <td>6.046789</td>\n","      <td>73819</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>72049be7bd30ea61297ea624ae198067</td>\n","      <td>219648</td>\n","      <td>4096</td>\n","      <td>6.497018</td>\n","      <td>219524</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>c9b3700a77facf29172f32df6bc77f48</td>\n","      <td>262144</td>\n","      <td>4096</td>\n","      <td>6.638142</td>\n","      <td>261943</td>\n","      <td>1</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["                               hash  size_of_data  ...  virtual_size  malware\n","0  071e8c3f8922e186e57548cd4c703a5d        443392  ...        442984        1\n","1  33f8e6d08a6aae939f25a8e0d63dd523        331264  ...        330784        1\n","2  b68abd064e975e1c6d5f25e748663076         74240  ...         73819        1\n","3  72049be7bd30ea61297ea624ae198067        219648  ...        219524        1\n","4  c9b3700a77facf29172f32df6bc77f48        262144  ...        261943        1\n","\n","[5 rows x 6 columns]"]},"metadata":{"tags":[]},"execution_count":3}]},{"cell_type":"code","metadata":{"id":"RH_fA2FBAhVK","colab_type":"code","outputId":"60b27eab-9ae2-4200-e096-49c3fd40ce44","executionInfo":{"status":"ok","timestamp":1592059561644,"user_tz":-480,"elapsed":12370,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":253}},"source":["# Import Data\n","import_df = pd.read_csv(\"https://drive.google.com/uc?id=1freGdbuJLDGK4JQ5IMLtw6fACo3NG1CK\")\n","import_df.head()"],"execution_count":4,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>hash</th>\n","      <th>GetProcAddress</th>\n","      <th>ExitProcess</th>\n","      <th>WriteFile</th>\n","      <th>GetLastError</th>\n","      <th>CloseHandle</th>\n","      <th>FreeLibrary</th>\n","      <th>Sleep</th>\n","      <th>GetStdHandle</th>\n","      <th>MultiByteToWideChar</th>\n","      <th>GetCurrentThreadId</th>\n","      <th>FindClose</th>\n","      <th>LeaveCriticalSection</th>\n","      <th>EnterCriticalSection</th>\n","      <th>VirtualAlloc</th>\n","      <th>DeleteCriticalSection</th>\n","      <th>WideCharToMultiByte</th>\n","      <th>RegCloseKey</th>\n","      <th>ReadFile</th>\n","      <th>SetFilePointer</th>\n","      <th>GetCurrentProcess</th>\n","      <th>GetACP</th>\n","      <th>UnhandledExceptionFilter</th>\n","      <th>RaiseException</th>\n","      <th>RtlUnwind</th>\n","      <th>TlsGetValue</th>\n","      <th>TlsSetValue</th>\n","      <th>GetModuleHandleW</th>\n","      <th>GetCPInfo</th>\n","      <th>GetCurrentProcessId</th>\n","      <th>VirtualFree</th>\n","      <th>GetDC</th>\n","      <th>GetModuleFileNameW</th>\n","      <th>CreateFileW</th>\n","      <th>SetEndOfFile</th>\n","      <th>SetLastError</th>\n","      <th>GetCommandLineW</th>\n","      <th>DestroyWindow</th>\n","      <th>QueryPerformanceCounter</th>\n","      <th>VirtualQuery</th>\n","      <th>...</th>\n","      <th>malloc</th>\n","      <th>_XcptFilter</th>\n","      <th>__vbaFreeVarList</th>\n","      <th>CoInitializeEx</th>\n","      <th>GetTextColor</th>\n","      <th>GetUserNameA</th>\n","      <th>listen</th>\n","      <th>inet_addr</th>\n","      <th>TextOutW</th>\n","      <th>__p__commode</th>\n","      <th>IsValidAcl</th>\n","      <th>gethostname</th>\n","      <th>PostThreadMessageW</th>\n","      <th>ReportEventW</th>\n","      <th>__vbaFreeObj</th>\n","      <th>CreateFileMappingA</th>\n","      <th>DragQueryFileW</th>\n","      <th>_onexit</th>\n","      <th>ImpersonateLoggedOnUser</th>\n","      <th>InitializeAcl</th>\n","      <th>MessageBoxIndirectW</th>\n","      <th>SendDlgItemMessageW</th>\n","      <th>__vbaVarMove</th>\n","      <th>GetDriveTypeA</th>\n","      <th>GetCurrentObject</th>\n","      <th>GetNearestPaletteIndex</th>\n","      <th>DosDateTimeToFileTime</th>\n","      <th>FreeConsole</th>\n","      <th>SetFileSecurityA</th>\n","      <th>DuplicateToken</th>\n","      <th>bind</th>\n","      <th>RegEnumKeyExA</th>\n","      <th>WinHttpOpen</th>\n","      <th>_controlfp</th>\n","      <th>WinExec</th>\n","      <th>GetSecurityDescriptorDacl</th>\n","      <th>FindFirstFreeAce</th>\n","      <th>GetTimeFormatW</th>\n","      <th>LookupAccountSidW</th>\n","      <th>malware</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>071e8c3f8922e186e57548cd4c703a5d</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>33f8e6d08a6aae939f25a8e0d63dd523</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>b68abd064e975e1c6d5f25e748663076</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>72049be7bd30ea61297ea624ae198067</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>c9b3700a77facf29172f32df6bc77f48</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","  </tbody>\n","</table>\n","<p>5 rows × 1002 columns</p>\n","</div>"],"text/plain":["                               hash  GetProcAddress  ...  LookupAccountSidW  malware\n","0  071e8c3f8922e186e57548cd4c703a5d               1  ...                  0        1\n","1  33f8e6d08a6aae939f25a8e0d63dd523               1  ...                  0        1\n","2  b68abd064e975e1c6d5f25e748663076               1  ...                  0        1\n","3  72049be7bd30ea61297ea624ae198067               1  ...                  0        1\n","4  c9b3700a77facf29172f32df6bc77f48               1  ...                  0        1\n","\n","[5 rows x 1002 columns]"]},"metadata":{"tags":[]},"execution_count":4}]},{"cell_type":"code","metadata":{"id":"FmZguYdXAvvH","colab_type":"code","outputId":"8810b6e8-72be-4313-f5c5-95da3ce8ecbf","executionInfo":{"status":"ok","timestamp":1592059566296,"user_tz":-480,"elapsed":16995,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":253}},"source":["# Image Data\n","image_df = pd.read_csv(\"/content/drive/My Drive/Colab Notebooks/dataset/raw_pe_images.csv\")\n","image_df.head()"],"execution_count":5,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>hash</th>\n","      <th>pix_0</th>\n","      <th>pix_1</th>\n","      <th>pix_2</th>\n","      <th>pix_3</th>\n","      <th>pix_4</th>\n","      <th>pix_5</th>\n","      <th>pix_6</th>\n","      <th>pix_7</th>\n","      <th>pix_8</th>\n","      <th>pix_9</th>\n","      <th>pix_10</th>\n","      <th>pix_11</th>\n","      <th>pix_12</th>\n","      <th>pix_13</th>\n","      <th>pix_14</th>\n","      <th>pix_15</th>\n","      <th>pix_16</th>\n","      <th>pix_17</th>\n","      <th>pix_18</th>\n","      <th>pix_19</th>\n","      <th>pix_20</th>\n","      <th>pix_21</th>\n","      <th>pix_22</th>\n","      <th>pix_23</th>\n","      <th>pix_24</th>\n","      <th>pix_25</th>\n","      <th>pix_26</th>\n","      <th>pix_27</th>\n","      <th>pix_28</th>\n","      <th>pix_29</th>\n","      <th>pix_30</th>\n","      <th>pix_31</th>\n","      <th>pix_32</th>\n","      <th>pix_33</th>\n","      <th>pix_34</th>\n","      <th>pix_35</th>\n","      <th>pix_36</th>\n","      <th>pix_37</th>\n","      <th>pix_38</th>\n","      <th>...</th>\n","      <th>pix_985</th>\n","      <th>pix_986</th>\n","      <th>pix_987</th>\n","      <th>pix_988</th>\n","      <th>pix_989</th>\n","      <th>pix_990</th>\n","      <th>pix_991</th>\n","      <th>pix_992</th>\n","      <th>pix_993</th>\n","      <th>pix_994</th>\n","      <th>pix_995</th>\n","      <th>pix_996</th>\n","      <th>pix_997</th>\n","      <th>pix_998</th>\n","      <th>pix_999</th>\n","      <th>pix_1000</th>\n","      <th>pix_1001</th>\n","      <th>pix_1002</th>\n","      <th>pix_1003</th>\n","      <th>pix_1004</th>\n","      <th>pix_1005</th>\n","      <th>pix_1006</th>\n","      <th>pix_1007</th>\n","      <th>pix_1008</th>\n","      <th>pix_1009</th>\n","      <th>pix_1010</th>\n","      <th>pix_1011</th>\n","      <th>pix_1012</th>\n","      <th>pix_1013</th>\n","      <th>pix_1014</th>\n","      <th>pix_1015</th>\n","      <th>pix_1016</th>\n","      <th>pix_1017</th>\n","      <th>pix_1018</th>\n","      <th>pix_1019</th>\n","      <th>pix_1020</th>\n","      <th>pix_1021</th>\n","      <th>pix_1022</th>\n","      <th>pix_1023</th>\n","      <th>malware</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>b324140e1fb35dc6b694879ba1f2be45</td>\n","      <td>15</td>\n","      <td>15</td>\n","      <td>239</td>\n","      <td>15</td>\n","      <td>223</td>\n","      <td>36</td>\n","      <td>102</td>\n","      <td>243</td>\n","      <td>102</td>\n","      <td>102</td>\n","      <td>254</td>\n","      <td>36</td>\n","      <td>40</td>\n","      <td>7</td>\n","      <td>102</td>\n","      <td>92</td>\n","      <td>15</td>\n","      <td>15</td>\n","      <td>102</td>\n","      <td>84</td>\n","      <td>239</td>\n","      <td>15</td>\n","      <td>15</td>\n","      <td>239</td>\n","      <td>192</td>\n","      <td>232</td>\n","      <td>64</td>\n","      <td>64</td>\n","      <td>3</td>\n","      <td>0</td>\n","      <td>229</td>\n","      <td>108</td>\n","      <td>0</td>\n","      <td>131</td>\n","      <td>83</td>\n","      <td>136</td>\n","      <td>76</td>\n","      <td>0</td>\n","      <td>255</td>\n","      <td>...</td>\n","      <td>252</td>\n","      <td>35</td>\n","      <td>194</td>\n","      <td>173</td>\n","      <td>219</td>\n","      <td>197</td>\n","      <td>254</td>\n","      <td>46</td>\n","      <td>101</td>\n","      <td>233</td>\n","      <td>209</td>\n","      <td>216</td>\n","      <td>32</td>\n","      <td>161</td>\n","      <td>4</td>\n","      <td>236</td>\n","      <td>39</td>\n","      <td>134</td>\n","      <td>188</td>\n","      <td>162</td>\n","      <td>108</td>\n","      <td>25</td>\n","      <td>189</td>\n","      <td>254</td>\n","      <td>114</td>\n","      <td>4</td>\n","      <td>238</td>\n","      <td>91</td>\n","      <td>66</td>\n","      <td>255</td>\n","      <td>31</td>\n","      <td>242</td>\n","      <td>245</td>\n","      <td>169</td>\n","      <td>212</td>\n","      <td>17</td>\n","      <td>255</td>\n","      <td>160</td>\n","      <td>122</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>1d32b1326a524b163eb74af645cd34d5</td>\n","      <td>234</td>\n","      <td>196</td>\n","      <td>8</td>\n","      <td>20</td>\n","      <td>182</td>\n","      <td>56</td>\n","      <td>27</td>\n","      <td>223</td>\n","      <td>144</td>\n","      <td>255</td>\n","      <td>207</td>\n","      <td>0</td>\n","      <td>77</td>\n","      <td>81</td>\n","      <td>112</td>\n","      <td>176</td>\n","      <td>131</td>\n","      <td>222</td>\n","      <td>100</td>\n","      <td>236</td>\n","      <td>83</td>\n","      <td>3</td>\n","      <td>4</td>\n","      <td>136</td>\n","      <td>136</td>\n","      <td>252</td>\n","      <td>235</td>\n","      <td>84</td>\n","      <td>30</td>\n","      <td>235</td>\n","      <td>118</td>\n","      <td>124</td>\n","      <td>224</td>\n","      <td>77</td>\n","      <td>27</td>\n","      <td>112</td>\n","      <td>125</td>\n","      <td>55</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>58</td>\n","      <td>44</td>\n","      <td>7</td>\n","      <td>124</td>\n","      <td>67</td>\n","      <td>90</td>\n","      <td>242</td>\n","      <td>212</td>\n","      <td>41</td>\n","      <td>228</td>\n","      <td>172</td>\n","      <td>193</td>\n","      <td>242</td>\n","      <td>131</td>\n","      <td>50</td>\n","      <td>9</td>\n","      <td>88</td>\n","      <td>253</td>\n","      <td>78</td>\n","      <td>207</td>\n","      <td>35</td>\n","      <td>249</td>\n","      <td>162</td>\n","      <td>162</td>\n","      <td>36</td>\n","      <td>220</td>\n","      <td>59</td>\n","      <td>180</td>\n","      <td>74</td>\n","      <td>213</td>\n","      <td>189</td>\n","      <td>228</td>\n","      <td>156</td>\n","      <td>54</td>\n","      <td>189</td>\n","      <td>82</td>\n","      <td>177</td>\n","      <td>125</td>\n","      <td>193</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>e44fea4913fc9fd91b8b07c4670aeac4</td>\n","      <td>196</td>\n","      <td>255</td>\n","      <td>5</td>\n","      <td>97</td>\n","      <td>35</td>\n","      <td>112</td>\n","      <td>219</td>\n","      <td>189</td>\n","      <td>217</td>\n","      <td>66</td>\n","      <td>36</td>\n","      <td>90</td>\n","      <td>117</td>\n","      <td>0</td>\n","      <td>69</td>\n","      <td>217</td>\n","      <td>132</td>\n","      <td>221</td>\n","      <td>0</td>\n","      <td>16</td>\n","      <td>65</td>\n","      <td>16</td>\n","      <td>139</td>\n","      <td>139</td>\n","      <td>255</td>\n","      <td>228</td>\n","      <td>203</td>\n","      <td>137</td>\n","      <td>69</td>\n","      <td>250</td>\n","      <td>229</td>\n","      <td>139</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>193</td>\n","      <td>230</td>\n","      <td>253</td>\n","      <td>197</td>\n","      <td>234</td>\n","      <td>255</td>\n","      <td>199</td>\n","      <td>196</td>\n","      <td>233</td>\n","      <td>254</td>\n","      <td>203</td>\n","      <td>237</td>\n","      <td>254</td>\n","      <td>205</td>\n","      <td>238</td>\n","      <td>255</td>\n","      <td>208</td>\n","      <td>241</td>\n","      <td>253</td>\n","      <td>211</td>\n","      <td>244</td>\n","      <td>255</td>\n","      <td>216</td>\n","      <td>245</td>\n","      <td>254</td>\n","      <td>219</td>\n","      <td>246</td>\n","      <td>255</td>\n","      <td>222</td>\n","      <td>224</td>\n","      <td>254</td>\n","      <td>189</td>\n","      <td>227</td>\n","      <td>255</td>\n","      <td>192</td>\n","      <td>231</td>\n","      <td>253</td>\n","      <td>196</td>\n","      <td>233</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>95badb16d862ba94ae85c44e4b31d749</td>\n","      <td>232</td>\n","      <td>252</td>\n","      <td>183</td>\n","      <td>39</td>\n","      <td>51</td>\n","      <td>1</td>\n","      <td>255</td>\n","      <td>87</td>\n","      <td>94</td>\n","      <td>128</td>\n","      <td>69</td>\n","      <td>252</td>\n","      <td>255</td>\n","      <td>236</td>\n","      <td>0</td>\n","      <td>150</td>\n","      <td>80</td>\n","      <td>116</td>\n","      <td>255</td>\n","      <td>100</td>\n","      <td>141</td>\n","      <td>198</td>\n","      <td>141</td>\n","      <td>43</td>\n","      <td>57</td>\n","      <td>1</td>\n","      <td>195</td>\n","      <td>3</td>\n","      <td>255</td>\n","      <td>235</td>\n","      <td>57</td>\n","      <td>232</td>\n","      <td>168</td>\n","      <td>1</td>\n","      <td>255</td>\n","      <td>137</td>\n","      <td>0</td>\n","      <td>241</td>\n","      <td>95</td>\n","      <td>...</td>\n","      <td>233</td>\n","      <td>226</td>\n","      <td>236</td>\n","      <td>177</td>\n","      <td>216</td>\n","      <td>233</td>\n","      <td>0</td>\n","      <td>92</td>\n","      <td>8</td>\n","      <td>59</td>\n","      <td>49</td>\n","      <td>1</td>\n","      <td>71</td>\n","      <td>57</td>\n","      <td>57</td>\n","      <td>220</td>\n","      <td>51</td>\n","      <td>56</td>\n","      <td>143</td>\n","      <td>90</td>\n","      <td>49</td>\n","      <td>59</td>\n","      <td>1</td>\n","      <td>203</td>\n","      <td>60</td>\n","      <td>48</td>\n","      <td>134</td>\n","      <td>138</td>\n","      <td>62</td>\n","      <td>48</td>\n","      <td>210</td>\n","      <td>52</td>\n","      <td>55</td>\n","      <td>95</td>\n","      <td>230</td>\n","      <td>48</td>\n","      <td>53</td>\n","      <td>103</td>\n","      <td>65</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>f30f32a4f42678ef49a543356d580232</td>\n","      <td>81</td>\n","      <td>84</td>\n","      <td>204</td>\n","      <td>228</td>\n","      <td>255</td>\n","      <td>157</td>\n","      <td>76</td>\n","      <td>254</td>\n","      <td>128</td>\n","      <td>39</td>\n","      <td>79</td>\n","      <td>255</td>\n","      <td>0</td>\n","      <td>255</td>\n","      <td>48</td>\n","      <td>0</td>\n","      <td>80</td>\n","      <td>0</td>\n","      <td>198</td>\n","      <td>89</td>\n","      <td>69</td>\n","      <td>141</td>\n","      <td>254</td>\n","      <td>0</td>\n","      <td>141</td>\n","      <td>69</td>\n","      <td>224</td>\n","      <td>83</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>186</td>\n","      <td>0</td>\n","      <td>69</td>\n","      <td>4</td>\n","      <td>199</td>\n","      <td>255</td>\n","      <td>79</td>\n","      <td>59</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>233</td>\n","      <td>0</td>\n","      <td>233</td>\n","      <td>236</td>\n","      <td>233</td>\n","      <td>60</td>\n","      <td>233</td>\n","      <td>53</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>50</td>\n","      <td>62</td>\n","      <td>145</td>\n","      <td>230</td>\n","      <td>244</td>\n","      <td>53</td>\n","      <td>55</td>\n","      <td>127</td>\n","      <td>24</td>\n","      <td>0</td>\n","      <td>48</td>\n","      <td>0</td>\n","      <td>230</td>\n","      <td>57</td>\n","      <td>52</td>\n","      <td>0</td>\n","      <td>203</td>\n","      <td>55</td>\n","      <td>50</td>\n","      <td>24</td>\n","      <td>69</td>\n","      <td>64</td>\n","      <td>53</td>\n","      <td>52</td>\n","      <td>216</td>\n","      <td>200</td>\n","      <td>55</td>\n","      <td>58</td>\n","      <td>1</td>\n","    </tr>\n","  </tbody>\n","</table>\n","<p>5 rows × 1026 columns</p>\n","</div>"],"text/plain":["                               hash  pix_0  pix_1  ...  pix_1022  pix_1023  malware\n","0  b324140e1fb35dc6b694879ba1f2be45     15     15  ...       160       122        1\n","1  1d32b1326a524b163eb74af645cd34d5    234    196  ...       125       193        1\n","2  e44fea4913fc9fd91b8b07c4670aeac4    196    255  ...       196       233        1\n","3  95badb16d862ba94ae85c44e4b31d749    232    252  ...       103        65        1\n","4  f30f32a4f42678ef49a543356d580232     81     84  ...        55        58        1\n","\n","[5 rows x 1026 columns]"]},"metadata":{"tags":[]},"execution_count":5}]},{"cell_type":"markdown","metadata":{"id":"Lou7Zp8SNK5u","colab_type":"text"},"source":["### Merge Dataframes"]},{"cell_type":"code","metadata":{"id":"YY9vhTssJ4PY","colab_type":"code","colab":{}},"source":["# Merge those three dataset with key=\"hash\"\n","train_df = pd.merge(header_df, import_df, on=\"hash\")\n","train_df = pd.merge(train_df, image_df, on=\"hash\")"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"ZmeRGyy9LEPT","colab_type":"code","outputId":"9f021335-e932-4020-ff0b-85ebbdadbcb4","executionInfo":{"status":"ok","timestamp":1592059567880,"user_tz":-480,"elapsed":18543,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":253}},"source":["# Drop all duplicated observations\n","train_df = train_df.drop_duplicates(subset=[\"hash\"], keep=False)\n","train_df.head()"],"execution_count":7,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>hash</th>\n","      <th>size_of_data</th>\n","      <th>virtual_address</th>\n","      <th>entropy</th>\n","      <th>virtual_size</th>\n","      <th>malware_x</th>\n","      <th>GetProcAddress</th>\n","      <th>ExitProcess</th>\n","      <th>WriteFile</th>\n","      <th>GetLastError</th>\n","      <th>CloseHandle</th>\n","      <th>FreeLibrary</th>\n","      <th>Sleep</th>\n","      <th>GetStdHandle</th>\n","      <th>MultiByteToWideChar</th>\n","      <th>GetCurrentThreadId</th>\n","      <th>FindClose</th>\n","      <th>LeaveCriticalSection</th>\n","      <th>EnterCriticalSection</th>\n","      <th>VirtualAlloc</th>\n","      <th>DeleteCriticalSection</th>\n","      <th>WideCharToMultiByte</th>\n","      <th>RegCloseKey</th>\n","      <th>ReadFile</th>\n","      <th>SetFilePointer</th>\n","      <th>GetCurrentProcess</th>\n","      <th>GetACP</th>\n","      <th>UnhandledExceptionFilter</th>\n","      <th>RaiseException</th>\n","      <th>RtlUnwind</th>\n","      <th>TlsGetValue</th>\n","      <th>TlsSetValue</th>\n","      <th>GetModuleHandleW</th>\n","      <th>GetCPInfo</th>\n","      <th>GetCurrentProcessId</th>\n","      <th>VirtualFree</th>\n","      <th>GetDC</th>\n","      <th>GetModuleFileNameW</th>\n","      <th>CreateFileW</th>\n","      <th>SetEndOfFile</th>\n","      <th>...</th>\n","      <th>pix_985</th>\n","      <th>pix_986</th>\n","      <th>pix_987</th>\n","      <th>pix_988</th>\n","      <th>pix_989</th>\n","      <th>pix_990</th>\n","      <th>pix_991</th>\n","      <th>pix_992</th>\n","      <th>pix_993</th>\n","      <th>pix_994</th>\n","      <th>pix_995</th>\n","      <th>pix_996</th>\n","      <th>pix_997</th>\n","      <th>pix_998</th>\n","      <th>pix_999</th>\n","      <th>pix_1000</th>\n","      <th>pix_1001</th>\n","      <th>pix_1002</th>\n","      <th>pix_1003</th>\n","      <th>pix_1004</th>\n","      <th>pix_1005</th>\n","      <th>pix_1006</th>\n","      <th>pix_1007</th>\n","      <th>pix_1008</th>\n","      <th>pix_1009</th>\n","      <th>pix_1010</th>\n","      <th>pix_1011</th>\n","      <th>pix_1012</th>\n","      <th>pix_1013</th>\n","      <th>pix_1014</th>\n","      <th>pix_1015</th>\n","      <th>pix_1016</th>\n","      <th>pix_1017</th>\n","      <th>pix_1018</th>\n","      <th>pix_1019</th>\n","      <th>pix_1020</th>\n","      <th>pix_1021</th>\n","      <th>pix_1022</th>\n","      <th>pix_1023</th>\n","      <th>malware</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>071e8c3f8922e186e57548cd4c703a5d</td>\n","      <td>443392</td>\n","      <td>4096</td>\n","      <td>6.529624</td>\n","      <td>442984</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>140</td>\n","      <td>61</td>\n","      <td>255</td>\n","      <td>148</td>\n","      <td>239</td>\n","      <td>86</td>\n","      <td>52</td>\n","      <td>123</td>\n","      <td>197</td>\n","      <td>125</td>\n","      <td>179</td>\n","      <td>208</td>\n","      <td>227</td>\n","      <td>98</td>\n","      <td>77</td>\n","      <td>78</td>\n","      <td>198</td>\n","      <td>232</td>\n","      <td>23</td>\n","      <td>213</td>\n","      <td>110</td>\n","      <td>56</td>\n","      <td>107</td>\n","      <td>91</td>\n","      <td>123</td>\n","      <td>145</td>\n","      <td>105</td>\n","      <td>169</td>\n","      <td>70</td>\n","      <td>47</td>\n","      <td>192</td>\n","      <td>190</td>\n","      <td>10</td>\n","      <td>15</td>\n","      <td>81</td>\n","      <td>104</td>\n","      <td>187</td>\n","      <td>121</td>\n","      <td>61</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>33f8e6d08a6aae939f25a8e0d63dd523</td>\n","      <td>331264</td>\n","      <td>4096</td>\n","      <td>6.604314</td>\n","      <td>330784</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>162</td>\n","      <td>140</td>\n","      <td>135</td>\n","      <td>62</td>\n","      <td>48</td>\n","      <td>52</td>\n","      <td>55</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>b68abd064e975e1c6d5f25e748663076</td>\n","      <td>74240</td>\n","      <td>4096</td>\n","      <td>6.046789</td>\n","      <td>73819</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>112</td>\n","      <td>108</td>\n","      <td>114</td>\n","      <td>0</td>\n","      <td>114</td>\n","      <td>110</td>\n","      <td>0</td>\n","      <td>25</td>\n","      <td>163</td>\n","      <td>40</td>\n","      <td>5</td>\n","      <td>74</td>\n","      <td>174</td>\n","      <td>170</td>\n","      <td>56</td>\n","      <td>56</td>\n","      <td>57</td>\n","      <td>57</td>\n","      <td>57</td>\n","      <td>58</td>\n","      <td>58</td>\n","      <td>59</td>\n","      <td>62</td>\n","      <td>239</td>\n","      <td>160</td>\n","      <td>81</td>\n","      <td>2</td>\n","      <td>179</td>\n","      <td>124</td>\n","      <td>159</td>\n","      <td>49</td>\n","      <td>50</td>\n","      <td>50</td>\n","      <td>51</td>\n","      <td>52</td>\n","      <td>52</td>\n","      <td>53</td>\n","      <td>54</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>72049be7bd30ea61297ea624ae198067</td>\n","      <td>219648</td>\n","      <td>4096</td>\n","      <td>6.497018</td>\n","      <td>219524</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>194</td>\n","      <td>231</td>\n","      <td>237</td>\n","      <td>204</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>194</td>\n","      <td>231</td>\n","      <td>235</td>\n","      <td>202</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>196</td>\n","      <td>230</td>\n","      <td>235</td>\n","      <td>201</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>217</td>\n","      <td>220</td>\n","      <td>199</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>c9b3700a77facf29172f32df6bc77f48</td>\n","      <td>262144</td>\n","      <td>4096</td>\n","      <td>6.638142</td>\n","      <td>261943</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>236</td>\n","      <td>216</td>\n","      <td>186</td>\n","      <td>236</td>\n","      <td>131</td>\n","      <td>233</td>\n","      <td>0</td>\n","      <td>62</td>\n","      <td>57</td>\n","      <td>63</td>\n","      <td>58</td>\n","      <td>48</td>\n","      <td>55</td>\n","      <td>61</td>\n","      <td>49</td>\n","      <td>51</td>\n","      <td>56</td>\n","      <td>61</td>\n","      <td>15</td>\n","      <td>6</td>\n","      <td>20</td>\n","      <td>0</td>\n","      <td>182</td>\n","      <td>132</td>\n","      <td>57</td>\n","      <td>104</td>\n","      <td>201</td>\n","      <td>85</td>\n","      <td>58</td>\n","      <td>59</td>\n","      <td>49</td>\n","      <td>54</td>\n","      <td>54</td>\n","      <td>55</td>\n","      <td>56</td>\n","      <td>57</td>\n","      <td>61</td>\n","      <td>63</td>\n","      <td>50</td>\n","      <td>1</td>\n","    </tr>\n","  </tbody>\n","</table>\n","<p>5 rows × 2032 columns</p>\n","</div>"],"text/plain":["                               hash  size_of_data  ...  pix_1023  malware\n","0  071e8c3f8922e186e57548cd4c703a5d        443392  ...        61        1\n","1  33f8e6d08a6aae939f25a8e0d63dd523        331264  ...         0        1\n","2  b68abd064e975e1c6d5f25e748663076         74240  ...        54        1\n","3  72049be7bd30ea61297ea624ae198067        219648  ...         0        1\n","4  c9b3700a77facf29172f32df6bc77f48        262144  ...        50        1\n","\n","[5 rows x 2032 columns]"]},"metadata":{"tags":[]},"execution_count":7}]},{"cell_type":"code","metadata":{"id":"5-es4ldlN0Z7","colab_type":"code","outputId":"2847ba65-5998-479d-fe5f-c6f2fa00b887","executionInfo":{"status":"ok","timestamp":1592059567881,"user_tz":-480,"elapsed":18518,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":51}},"source":["# Total amount of data\n","print(\"There are {} columns in the training data now.\".format(len(train_df.columns)))\n","print(\"There are {} rows in the training data now.\".format(len(train_df)))"],"execution_count":8,"outputs":[{"output_type":"stream","text":["There are 2032 columns in the training data now.\n","There are 42283 rows in the training data now.\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"hCbbCOiW66Ow","colab_type":"code","outputId":"2ae8b509-79b0-422b-d711-4111bf736a9a","executionInfo":{"status":"ok","timestamp":1592059567882,"user_tz":-480,"elapsed":18497,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":204}},"source":["# check if observations are mislabeled\n","check_df = train_df[[\"malware_x\", \"malware_y\", \"malware\"]]\n","check_df.head()"],"execution_count":9,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>malware_x</th>\n","      <th>malware_y</th>\n","      <th>malware</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["   malware_x  malware_y  malware\n","0          1          1        1\n","1          1          1        1\n","2          1          1        1\n","3          1          1        1\n","4          1          1        1"]},"metadata":{"tags":[]},"execution_count":9}]},{"cell_type":"code","metadata":{"id":"-VD5-fiR8uWM","colab_type":"code","outputId":"2265e9f8-202b-43d3-8735-32872379f78b","executionInfo":{"status":"ok","timestamp":1592059568735,"user_tz":-480,"elapsed":19333,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["s = 0\n","for i in range(len(check_df)):\n","    if not (check_df.iloc[i,0] == check_df.iloc[i,1] == check_df.iloc[i,2]):\n","        print(False)\n","        s += 1\n","print(\"There are {} rows have been error labeled.\".format(s)) # correct      "],"execution_count":10,"outputs":[{"output_type":"stream","text":["There are 0 rows have been error labeled.\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"zyIUyvSJ-ZsD","colab_type":"code","outputId":"fccd510f-e19b-4809-c3ca-cc4c1f184eee","executionInfo":{"status":"ok","timestamp":1592059568736,"user_tz":-480,"elapsed":19316,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["# drop redundant columns\n","train_df = train_df.drop([\"malware_x\", \"malware_y\"], axis=1)\n","print(\"There are {} columns in the training data now.\".format(len(train_df.columns)))"],"execution_count":11,"outputs":[{"output_type":"stream","text":["There are 2030 columns in the training data now.\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"Lle4bCl7_Y8F","colab_type":"code","outputId":"50deef9f-1c53-4413-e775-3b5824626b84","executionInfo":{"status":"ok","timestamp":1592059568736,"user_tz":-480,"elapsed":19296,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":253}},"source":["train_df.head()"],"execution_count":12,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>hash</th>\n","      <th>size_of_data</th>\n","      <th>virtual_address</th>\n","      <th>entropy</th>\n","      <th>virtual_size</th>\n","      <th>GetProcAddress</th>\n","      <th>ExitProcess</th>\n","      <th>WriteFile</th>\n","      <th>GetLastError</th>\n","      <th>CloseHandle</th>\n","      <th>FreeLibrary</th>\n","      <th>Sleep</th>\n","      <th>GetStdHandle</th>\n","      <th>MultiByteToWideChar</th>\n","      <th>GetCurrentThreadId</th>\n","      <th>FindClose</th>\n","      <th>LeaveCriticalSection</th>\n","      <th>EnterCriticalSection</th>\n","      <th>VirtualAlloc</th>\n","      <th>DeleteCriticalSection</th>\n","      <th>WideCharToMultiByte</th>\n","      <th>RegCloseKey</th>\n","      <th>ReadFile</th>\n","      <th>SetFilePointer</th>\n","      <th>GetCurrentProcess</th>\n","      <th>GetACP</th>\n","      <th>UnhandledExceptionFilter</th>\n","      <th>RaiseException</th>\n","      <th>RtlUnwind</th>\n","      <th>TlsGetValue</th>\n","      <th>TlsSetValue</th>\n","      <th>GetModuleHandleW</th>\n","      <th>GetCPInfo</th>\n","      <th>GetCurrentProcessId</th>\n","      <th>VirtualFree</th>\n","      <th>GetDC</th>\n","      <th>GetModuleFileNameW</th>\n","      <th>CreateFileW</th>\n","      <th>SetEndOfFile</th>\n","      <th>SetLastError</th>\n","      <th>...</th>\n","      <th>pix_985</th>\n","      <th>pix_986</th>\n","      <th>pix_987</th>\n","      <th>pix_988</th>\n","      <th>pix_989</th>\n","      <th>pix_990</th>\n","      <th>pix_991</th>\n","      <th>pix_992</th>\n","      <th>pix_993</th>\n","      <th>pix_994</th>\n","      <th>pix_995</th>\n","      <th>pix_996</th>\n","      <th>pix_997</th>\n","      <th>pix_998</th>\n","      <th>pix_999</th>\n","      <th>pix_1000</th>\n","      <th>pix_1001</th>\n","      <th>pix_1002</th>\n","      <th>pix_1003</th>\n","      <th>pix_1004</th>\n","      <th>pix_1005</th>\n","      <th>pix_1006</th>\n","      <th>pix_1007</th>\n","      <th>pix_1008</th>\n","      <th>pix_1009</th>\n","      <th>pix_1010</th>\n","      <th>pix_1011</th>\n","      <th>pix_1012</th>\n","      <th>pix_1013</th>\n","      <th>pix_1014</th>\n","      <th>pix_1015</th>\n","      <th>pix_1016</th>\n","      <th>pix_1017</th>\n","      <th>pix_1018</th>\n","      <th>pix_1019</th>\n","      <th>pix_1020</th>\n","      <th>pix_1021</th>\n","      <th>pix_1022</th>\n","      <th>pix_1023</th>\n","      <th>malware</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>071e8c3f8922e186e57548cd4c703a5d</td>\n","      <td>443392</td>\n","      <td>4096</td>\n","      <td>6.529624</td>\n","      <td>442984</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>140</td>\n","      <td>61</td>\n","      <td>255</td>\n","      <td>148</td>\n","      <td>239</td>\n","      <td>86</td>\n","      <td>52</td>\n","      <td>123</td>\n","      <td>197</td>\n","      <td>125</td>\n","      <td>179</td>\n","      <td>208</td>\n","      <td>227</td>\n","      <td>98</td>\n","      <td>77</td>\n","      <td>78</td>\n","      <td>198</td>\n","      <td>232</td>\n","      <td>23</td>\n","      <td>213</td>\n","      <td>110</td>\n","      <td>56</td>\n","      <td>107</td>\n","      <td>91</td>\n","      <td>123</td>\n","      <td>145</td>\n","      <td>105</td>\n","      <td>169</td>\n","      <td>70</td>\n","      <td>47</td>\n","      <td>192</td>\n","      <td>190</td>\n","      <td>10</td>\n","      <td>15</td>\n","      <td>81</td>\n","      <td>104</td>\n","      <td>187</td>\n","      <td>121</td>\n","      <td>61</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>33f8e6d08a6aae939f25a8e0d63dd523</td>\n","      <td>331264</td>\n","      <td>4096</td>\n","      <td>6.604314</td>\n","      <td>330784</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>162</td>\n","      <td>140</td>\n","      <td>135</td>\n","      <td>62</td>\n","      <td>48</td>\n","      <td>52</td>\n","      <td>55</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>b68abd064e975e1c6d5f25e748663076</td>\n","      <td>74240</td>\n","      <td>4096</td>\n","      <td>6.046789</td>\n","      <td>73819</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>112</td>\n","      <td>108</td>\n","      <td>114</td>\n","      <td>0</td>\n","      <td>114</td>\n","      <td>110</td>\n","      <td>0</td>\n","      <td>25</td>\n","      <td>163</td>\n","      <td>40</td>\n","      <td>5</td>\n","      <td>74</td>\n","      <td>174</td>\n","      <td>170</td>\n","      <td>56</td>\n","      <td>56</td>\n","      <td>57</td>\n","      <td>57</td>\n","      <td>57</td>\n","      <td>58</td>\n","      <td>58</td>\n","      <td>59</td>\n","      <td>62</td>\n","      <td>239</td>\n","      <td>160</td>\n","      <td>81</td>\n","      <td>2</td>\n","      <td>179</td>\n","      <td>124</td>\n","      <td>159</td>\n","      <td>49</td>\n","      <td>50</td>\n","      <td>50</td>\n","      <td>51</td>\n","      <td>52</td>\n","      <td>52</td>\n","      <td>53</td>\n","      <td>54</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>72049be7bd30ea61297ea624ae198067</td>\n","      <td>219648</td>\n","      <td>4096</td>\n","      <td>6.497018</td>\n","      <td>219524</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>194</td>\n","      <td>231</td>\n","      <td>237</td>\n","      <td>204</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>194</td>\n","      <td>231</td>\n","      <td>235</td>\n","      <td>202</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>196</td>\n","      <td>230</td>\n","      <td>235</td>\n","      <td>201</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>217</td>\n","      <td>220</td>\n","      <td>199</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>c9b3700a77facf29172f32df6bc77f48</td>\n","      <td>262144</td>\n","      <td>4096</td>\n","      <td>6.638142</td>\n","      <td>261943</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>236</td>\n","      <td>216</td>\n","      <td>186</td>\n","      <td>236</td>\n","      <td>131</td>\n","      <td>233</td>\n","      <td>0</td>\n","      <td>62</td>\n","      <td>57</td>\n","      <td>63</td>\n","      <td>58</td>\n","      <td>48</td>\n","      <td>55</td>\n","      <td>61</td>\n","      <td>49</td>\n","      <td>51</td>\n","      <td>56</td>\n","      <td>61</td>\n","      <td>15</td>\n","      <td>6</td>\n","      <td>20</td>\n","      <td>0</td>\n","      <td>182</td>\n","      <td>132</td>\n","      <td>57</td>\n","      <td>104</td>\n","      <td>201</td>\n","      <td>85</td>\n","      <td>58</td>\n","      <td>59</td>\n","      <td>49</td>\n","      <td>54</td>\n","      <td>54</td>\n","      <td>55</td>\n","      <td>56</td>\n","      <td>57</td>\n","      <td>61</td>\n","      <td>63</td>\n","      <td>50</td>\n","      <td>1</td>\n","    </tr>\n","  </tbody>\n","</table>\n","<p>5 rows × 2030 columns</p>\n","</div>"],"text/plain":["                               hash  size_of_data  ...  pix_1023  malware\n","0  071e8c3f8922e186e57548cd4c703a5d        443392  ...        61        1\n","1  33f8e6d08a6aae939f25a8e0d63dd523        331264  ...         0        1\n","2  b68abd064e975e1c6d5f25e748663076         74240  ...        54        1\n","3  72049be7bd30ea61297ea624ae198067        219648  ...         0        1\n","4  c9b3700a77facf29172f32df6bc77f48        262144  ...        50        1\n","\n","[5 rows x 2030 columns]"]},"metadata":{"tags":[]},"execution_count":12}]},{"cell_type":"markdown","metadata":{"id":"g2PmYcgvNRfe","colab_type":"text"},"source":["### Feature Engineering"]},{"cell_type":"code","metadata":{"id":"1P9kCMPGNVUC","colab_type":"code","outputId":"e3326742-0d3b-4e24-da82-e0a62762281b","executionInfo":{"status":"ok","timestamp":1592059568736,"user_tz":-480,"elapsed":19275,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":253}},"source":["# Create new feature called \"disk_memory_ratio\"\n","disk_memory_ratio = train_df[\"size_of_data\"] / train_df[\"virtual_size\"]\n","train_df.insert(loc=5, column=\"disk_memory_ratio\", value=disk_memory_ratio)\n","train_df.head()"],"execution_count":13,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>hash</th>\n","      <th>size_of_data</th>\n","      <th>virtual_address</th>\n","      <th>entropy</th>\n","      <th>virtual_size</th>\n","      <th>disk_memory_ratio</th>\n","      <th>GetProcAddress</th>\n","      <th>ExitProcess</th>\n","      <th>WriteFile</th>\n","      <th>GetLastError</th>\n","      <th>CloseHandle</th>\n","      <th>FreeLibrary</th>\n","      <th>Sleep</th>\n","      <th>GetStdHandle</th>\n","      <th>MultiByteToWideChar</th>\n","      <th>GetCurrentThreadId</th>\n","      <th>FindClose</th>\n","      <th>LeaveCriticalSection</th>\n","      <th>EnterCriticalSection</th>\n","      <th>VirtualAlloc</th>\n","      <th>DeleteCriticalSection</th>\n","      <th>WideCharToMultiByte</th>\n","      <th>RegCloseKey</th>\n","      <th>ReadFile</th>\n","      <th>SetFilePointer</th>\n","      <th>GetCurrentProcess</th>\n","      <th>GetACP</th>\n","      <th>UnhandledExceptionFilter</th>\n","      <th>RaiseException</th>\n","      <th>RtlUnwind</th>\n","      <th>TlsGetValue</th>\n","      <th>TlsSetValue</th>\n","      <th>GetModuleHandleW</th>\n","      <th>GetCPInfo</th>\n","      <th>GetCurrentProcessId</th>\n","      <th>VirtualFree</th>\n","      <th>GetDC</th>\n","      <th>GetModuleFileNameW</th>\n","      <th>CreateFileW</th>\n","      <th>SetEndOfFile</th>\n","      <th>...</th>\n","      <th>pix_985</th>\n","      <th>pix_986</th>\n","      <th>pix_987</th>\n","      <th>pix_988</th>\n","      <th>pix_989</th>\n","      <th>pix_990</th>\n","      <th>pix_991</th>\n","      <th>pix_992</th>\n","      <th>pix_993</th>\n","      <th>pix_994</th>\n","      <th>pix_995</th>\n","      <th>pix_996</th>\n","      <th>pix_997</th>\n","      <th>pix_998</th>\n","      <th>pix_999</th>\n","      <th>pix_1000</th>\n","      <th>pix_1001</th>\n","      <th>pix_1002</th>\n","      <th>pix_1003</th>\n","      <th>pix_1004</th>\n","      <th>pix_1005</th>\n","      <th>pix_1006</th>\n","      <th>pix_1007</th>\n","      <th>pix_1008</th>\n","      <th>pix_1009</th>\n","      <th>pix_1010</th>\n","      <th>pix_1011</th>\n","      <th>pix_1012</th>\n","      <th>pix_1013</th>\n","      <th>pix_1014</th>\n","      <th>pix_1015</th>\n","      <th>pix_1016</th>\n","      <th>pix_1017</th>\n","      <th>pix_1018</th>\n","      <th>pix_1019</th>\n","      <th>pix_1020</th>\n","      <th>pix_1021</th>\n","      <th>pix_1022</th>\n","      <th>pix_1023</th>\n","      <th>malware</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>071e8c3f8922e186e57548cd4c703a5d</td>\n","      <td>443392</td>\n","      <td>4096</td>\n","      <td>6.529624</td>\n","      <td>442984</td>\n","      <td>1.000921</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>140</td>\n","      <td>61</td>\n","      <td>255</td>\n","      <td>148</td>\n","      <td>239</td>\n","      <td>86</td>\n","      <td>52</td>\n","      <td>123</td>\n","      <td>197</td>\n","      <td>125</td>\n","      <td>179</td>\n","      <td>208</td>\n","      <td>227</td>\n","      <td>98</td>\n","      <td>77</td>\n","      <td>78</td>\n","      <td>198</td>\n","      <td>232</td>\n","      <td>23</td>\n","      <td>213</td>\n","      <td>110</td>\n","      <td>56</td>\n","      <td>107</td>\n","      <td>91</td>\n","      <td>123</td>\n","      <td>145</td>\n","      <td>105</td>\n","      <td>169</td>\n","      <td>70</td>\n","      <td>47</td>\n","      <td>192</td>\n","      <td>190</td>\n","      <td>10</td>\n","      <td>15</td>\n","      <td>81</td>\n","      <td>104</td>\n","      <td>187</td>\n","      <td>121</td>\n","      <td>61</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>33f8e6d08a6aae939f25a8e0d63dd523</td>\n","      <td>331264</td>\n","      <td>4096</td>\n","      <td>6.604314</td>\n","      <td>330784</td>\n","      <td>1.001451</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>162</td>\n","      <td>140</td>\n","      <td>135</td>\n","      <td>62</td>\n","      <td>48</td>\n","      <td>52</td>\n","      <td>55</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>2</th>\n","      <td>b68abd064e975e1c6d5f25e748663076</td>\n","      <td>74240</td>\n","      <td>4096</td>\n","      <td>6.046789</td>\n","      <td>73819</td>\n","      <td>1.005703</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>0</td>\n","      <td>112</td>\n","      <td>108</td>\n","      <td>114</td>\n","      <td>0</td>\n","      <td>114</td>\n","      <td>110</td>\n","      <td>0</td>\n","      <td>25</td>\n","      <td>163</td>\n","      <td>40</td>\n","      <td>5</td>\n","      <td>74</td>\n","      <td>174</td>\n","      <td>170</td>\n","      <td>56</td>\n","      <td>56</td>\n","      <td>57</td>\n","      <td>57</td>\n","      <td>57</td>\n","      <td>58</td>\n","      <td>58</td>\n","      <td>59</td>\n","      <td>62</td>\n","      <td>239</td>\n","      <td>160</td>\n","      <td>81</td>\n","      <td>2</td>\n","      <td>179</td>\n","      <td>124</td>\n","      <td>159</td>\n","      <td>49</td>\n","      <td>50</td>\n","      <td>50</td>\n","      <td>51</td>\n","      <td>52</td>\n","      <td>52</td>\n","      <td>53</td>\n","      <td>54</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>3</th>\n","      <td>72049be7bd30ea61297ea624ae198067</td>\n","      <td>219648</td>\n","      <td>4096</td>\n","      <td>6.497018</td>\n","      <td>219524</td>\n","      <td>1.000565</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>...</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>255</td>\n","      <td>194</td>\n","      <td>231</td>\n","      <td>237</td>\n","      <td>204</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>194</td>\n","      <td>231</td>\n","      <td>235</td>\n","      <td>202</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>196</td>\n","      <td>230</td>\n","      <td>235</td>\n","      <td>201</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>217</td>\n","      <td>220</td>\n","      <td>199</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>0</td>\n","      <td>1</td>\n","    </tr>\n","    <tr>\n","      <th>4</th>\n","      <td>c9b3700a77facf29172f32df6bc77f48</td>\n","      <td>262144</td>\n","      <td>4096</td>\n","      <td>6.638142</td>\n","      <td>261943</td>\n","      <td>1.000767</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>1</td>\n","      <td>...</td>\n","      <td>236</td>\n","      <td>216</td>\n","      <td>186</td>\n","      <td>236</td>\n","      <td>131</td>\n","      <td>233</td>\n","      <td>0</td>\n","      <td>62</td>\n","      <td>57</td>\n","      <td>63</td>\n","      <td>58</td>\n","      <td>48</td>\n","      <td>55</td>\n","      <td>61</td>\n","      <td>49</td>\n","      <td>51</td>\n","      <td>56</td>\n","      <td>61</td>\n","      <td>15</td>\n","      <td>6</td>\n","      <td>20</td>\n","      <td>0</td>\n","      <td>182</td>\n","      <td>132</td>\n","      <td>57</td>\n","      <td>104</td>\n","      <td>201</td>\n","      <td>85</td>\n","      <td>58</td>\n","      <td>59</td>\n","      <td>49</td>\n","      <td>54</td>\n","      <td>54</td>\n","      <td>55</td>\n","      <td>56</td>\n","      <td>57</td>\n","      <td>61</td>\n","      <td>63</td>\n","      <td>50</td>\n","      <td>1</td>\n","    </tr>\n","  </tbody>\n","</table>\n","<p>5 rows × 2031 columns</p>\n","</div>"],"text/plain":["                               hash  size_of_data  ...  pix_1023  malware\n","0  071e8c3f8922e186e57548cd4c703a5d        443392  ...        61        1\n","1  33f8e6d08a6aae939f25a8e0d63dd523        331264  ...         0        1\n","2  b68abd064e975e1c6d5f25e748663076         74240  ...        54        1\n","3  72049be7bd30ea61297ea624ae198067        219648  ...         0        1\n","4  c9b3700a77facf29172f32df6bc77f48        262144  ...        50        1\n","\n","[5 rows x 2031 columns]"]},"metadata":{"tags":[]},"execution_count":13}]},{"cell_type":"markdown","metadata":{"id":"yclpoV2XxYAA","colab_type":"text"},"source":["### EDA"]},{"cell_type":"code","metadata":{"id":"DvwQy6flYFqN","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":285},"outputId":"02848eec-0eb8-48e2-c168-69ce79ecd607","executionInfo":{"status":"ok","timestamp":1592059569282,"user_tz":-480,"elapsed":19801,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}}},"source":["# malware vs. benign distribution\n","# we need resampling\n","sns.set()\n","plt.bar([0,1], [train_df.malware.value_counts()[0], train_df.malware.value_counts()[1]])\n","plt.xticks([0,1])\n","plt.xlabel(\"malware (0: benign / 1:malware)\")\n","plt.ylabel(\"count\")\n","plt.show()"],"execution_count":14,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"grOfBYADuXUJ","colab_type":"code","outputId":"f36b7be6-4c10-4836-f54c-b66b98f38969","executionInfo":{"status":"ok","timestamp":1592059569283,"user_tz":-480,"elapsed":19783,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":285}},"source":["# entropy distribution\n","plt.hist(train_df.entropy)\n","plt.xlabel(\"entropy\")\n","plt.ylabel(\"count\")\n","plt.show()"],"execution_count":15,"outputs":[{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAZoAAAEMCAYAAAD9OXA9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAcBElEQVR4nO3df1DUdeLH8dcutCiKIgi6oN8smxgupyiYrBnNC85TE8W7rtNhrOvMfmhylql5adCh1oHWVYbSD6cfM07e3JQgRGEdNZX9OJ1qjLB0zEplEwW8RBFj9/P9g3GTK2sX9r0fxOfjL3ff+9nPa5foxef9+ex7HZZlWQIAwBCn3QEAAL0bRQMAMIqiAQAYRdEAAIyiaAAARlE0AACjKBoAgFGRdgfoqZqbj8nnC/4jRvHx/dXY2GIgUfeQKzjkCg65gtMbczmdDg0a1O8nxyiaM/D5rC4VzalteyJyBYdcwSFXcM6lXEydAQCMomgAAEZRNAAAo8JWNHPnztXUqVM1bdo05ebmaufOnZKkvXv3avr06ZowYYKmT5+ur776yr+NiTEAQHiFrWiKioq0efNmlZWVadasWbrvvvskSQUFBcrNzVV1dbVyc3OVn5/v38bEGAAgvMJWNDExMf5/t7S0yOFwqLGxUXV1dcrOzpYkZWdnq66uTk1NTUbGAADhF9bLm5cuXaqtW7fKsiw988wz8ng8GjJkiCIiIiRJERERSkxMlMfjkWVZIR+Li4sL58sFACjMRbNy5UpJUllZmYqLizV//vxw7j4o8fH9u7xtQkLMLz/IBuQKDrmC09tynfzeK9d5ESFO84Ofy2V63z/HxM/Rlg9sTps2Tfn5+Ro6dKgOHjwor9eriIgIeb1eNTQ0yO12y7KskI8Fo7GxpUsfXEpIiNGhQ0eD3s40cgWHXMHpjbkSEmI05Z7yECcKTMXDOba8n915v5xOxxn/QA/LOZpjx47J4/H4b9fU1GjgwIGKj49XamqqKisrJUmVlZVKTU1VXFyckTEAQPiF5YimtbVV8+fPV2trq5xOpwYOHKjS0lI5HA498MADWrJkidauXasBAwaoqKjIv52JMQBAeDksy+qZC+7YjKmz8CBXcMgVHKbOgnNWT50BAM5dFA0AwCiKBgBgFEUDADCKogEAGEXRAACMomgAAEZRNAAAoygaAIBRFA0AwCiKBgBgFEUDADCKogEAGEXRAACMomgAAEZRNAAAoygaAIBRFA0AwCiKBgBgFEUDADCKogEAGEXRAACMomgAAEZRNAAAoygaAIBRFA0AwKjIcOykublZixcv1jfffCOXy6Xzzz9fhYWFiouLU0pKii6++GI5nR2dV1xcrJSUFElSTU2NiouL5fV6dckll+ihhx5S3759uzUGAAivsBzROBwOzZ49W9XV1aqoqNDw4cO1evVq//jGjRtVXl6u8vJyf8kcO3ZM999/v0pLS/X666+rX79+Wr9+fbfGAADhF5aiiY2N1ejRo/2309LSVF9f/7PbvP322xo1apRGjBghSZoxY4ZeffXVbo0BAMIvLFNnp/P5fHrxxReVmZnpv+/GG2+U1+vVNddco7y8PLlcLnk8HiUlJfkfk5SUJI/HI0ldHgtGfHz/oLc5JSEhpsvbmkSu4JArOOQKLbtym9hv2Itm+fLlio6O1syZMyVJb731ltxut1paWrRo0SKVlJTo7rvvDnesH2lsbJHPZwW9XUJCjA4dOmogUfeQKzjkCk5vzGV3Qdnxfnbn/XI6HWf8Az2sV50VFRXp66+/1qOPPuo/+e92uyVJ/fv31w033KCPPvrIf//p02v19fX+x3Z1DAAQfmErmkceeUS1tbUqKSmRy+WSJP33v//ViRMnJEnt7e2qrq5WamqqJGns2LH69NNP9dVXX0nquGBg0qRJ3RoDAIRfWKbOdu/erSeffFIjRozQjBkzJEnDhg3T7NmzlZ+fL4fDofb2dl1++eWaP3++pI4jnMLCQt1+++3y+XxKTU3V0qVLuzUGAAg/h2VZwZ+IOAdwjiY8yBUccgWnu+doptxTHuJEgal4OIdzNAAABIqiAQAYRdEAAIyiaAAARlE0AACjKBoAgFEUDQDAKIoGAGAURQMAMIqiAQAYRdEAAIyiaAAARlE0AACjKBoAgFEUDQDAKIoGAGAURQMAMIqiAQAYRdEAAIyiaAAARlE0AACjKBoAgFEUDQDAKIoGAGAURQMAMCosRdPc3Kxbb71VEyZM0JQpUzRv3jw1NTVJkj755BNNnTpVEyZM0KxZs9TY2OjfzsQYACC8wlI0DodDs2fPVnV1tSoqKjR8+HCtXr1aPp9PixYtUn5+vqqrq5WRkaHVq1dLkpExAED4haVoYmNjNXr0aP/ttLQ01dfXq7a2VlFRUcrIyJAkzZgxQ6+99pokGRkDAIRf2M/R+Hw+vfjii8rMzJTH41FSUpJ/LC4uTj6fT0eOHDEyBgAIv8hw73D58uWKjo7WzJkz9frrr4d79wGLj+/f5W0TEmJCmCR0yBUccgWHXKFlV24T+w1r0RQVFenrr79WaWmpnE6n3G636uvr/eNNTU1yOp2KjY01MhaMxsYW+XxW0K8xISFGhw4dDXo708gVHHIFpzfmsrug7Hg/u/N+OZ2OM/6BHraps0ceeUS1tbUqKSmRy+WSJI0aNUonTpzQ9u3bJUkbN27UxIkTjY0BAMIvLEc0u3fv1pNPPqkRI0ZoxowZkqRhw4appKRExcXFKigoUFtbm5KTk7Vq1SpJktPpDPkYACD8HJZlBT8/dA5g6iw8yBUccgWnu1NnU+4pD3GiwFQ8nMPUGQAAgaJoAABGUTQAAKMoGgCAURQNAMAoigYAYBRFAwAwiqIBABhF0QAAjKJoAABGBVw069ev/8n7n3322ZCFAQD0PgEXTUlJyU/ev27dupCFAQD0Pr+4evP7778vqeObMT/44AOdvgbn/v371a9fP3PpAABnvV8smqVLl0qS2tradN999/nvdzgcSkhI0LJly8ylAwCc9X6xaGpqaiRJixcvVnFxsfFAAIDeJeAvPju9ZHw+X6cxp5OL1wAAPy3govnss89UWFioL774Qm1tbZIky7LkcDi0c+dOYwEBAGe3gItmyZIluvbaa/Xggw+qT58+JjMBAHqRgIvmwIEDuvvuu+VwOEzmAQD0MgGfXBk/frzeffddk1kAAL1QwEc0bW1tmjdvntLT0zV48OBOY1yNBgA4k4CL5qKLLtJFF11kMgsAoBcKuGjmzZtnMgcAoJcKuGhOLUXzU66++uqQhAEA9D4BF82ppWhOaW5u1vfff68hQ4bo3//+d8iDAQB6h4CL5tRSNKd4vV6tW7eORTUBAD+ry2vHRERE6I477tAzzzwT0OOLioqUmZmplJQU7dq1y39/ZmamJk6cqJycHOXk5Oidd97xj33yySeaOnWqJkyYoFmzZqmxsbHbYwCA8OrWImVbt24N+AOcWVlZ2rBhg5KTk3809vjjj6u8vFzl5eUaO3aspI711BYtWqT8/HxVV1crIyNDq1ev7tYYACD8Ap46GzduXKdSaW1t1cmTJ1VQUBDQ9hkZGUEFq62tVVRUlH+7GTNmKCsrSw899FCXxwAA4Rdw0axatarT7b59++qCCy5Q//79ux1i4cKFsixL6enpWrBggQYMGCCPx6OkpCT/Y+Li4uTz+XTkyJEuj8XGxnY7KwAgOAEXzZVXXimpY2rq8OHDGjx4cEi+HmDDhg1yu906efKkVq5cqcLCwh4x1RUf3/UCTUiICWGS0CFXcMgVHHKFll25Tew34KJpaWlRYWGhqqqq1N7ersjISE2ePFnLli1TTEzXg7ndbkmSy+VSbm6u5syZ47+/vr7e/7impiY5nU7FxsZ2eSwYjY0t8vmsX37g/0hIiNGhQ0eD3s40cgWHXMHpjbnsLig73s/uvF9Op+OMf6AHfEiyYsUKtba2qqKiQjt27FBFRYVaW1u1YsWKLoWSpOPHj+vo0Y4XZVmWqqqqlJqaKkkaNWqUTpw4oe3bt0uSNm7cqIkTJ3ZrDAAQfgEf0bzzzjt644031LdvX0nSBRdcoIceekjjx48PaPsVK1Zoy5YtOnz4sP785z8rNjZWpaWlysvLk9frlc/n08iRI/0XFzidThUXF6ugoEBtbW1KTk72nyfq6hgAIPwCLpqoqCg1NTV1ujy5ublZLpcroO2XLVumZcuW/ej+srKyM25zxRVXqKKiIqRjAIDwCrho/vCHP2jWrFm6+eablZSUpPr6ej333HO64YYbTOYDAJzlAi6aOXPmaMiQIaqoqFBDQ4MSExM1e/ZsigYA8LMCvhhg5cqVuuCCC/Tcc8+pqqpKzz33nEaOHKmVK1eazAcAOMsFXDSVlZUaNWpUp/tGjRqlysrKkIcCAPQeAReNw+GQz+frdN+pq8UAADiTgIsmIyNDjz32mL9YfD6f1qxZE/QaZgCAc0tQX3x2++23a8yYMUpKSpLH41FCQoJKS0tN5gMAnOUCLpqhQ4dq06ZN2rFjhzwej9xuty699NKQrHcGAOi9Ai4aqeNT92lpaUpLSzOVBwDQy3A4AgAwiqIBABhF0QAAjKJoAABGUTQAAKMoGgCAURQNAMAoigYAYBRFAwAwiqIBABhF0QAAjKJoAABGUTQAAKMoGgCAURQNAMAoigYAYFRYiqaoqEiZmZlKSUnRrl27/Pfv3btX06dP14QJEzR9+nR99dVXRscAAOEXlqLJysrShg0blJyc3On+goIC5ebmqrq6Wrm5ucrPzzc6BgAIv7AUTUZGhtxud6f7GhsbVVdXp+zsbElSdna26urq1NTUZGQMAGCPSLt27PF4NGTIEEVEREiSIiIilJiYKI/HI8uyQj4WFxdnzwsFgHOcbUXT08XH9+/ytgkJMSFMEjrkCg65gkOu0LIrt4n92lY0brdbBw8elNfrVUREhLxerxoaGuR2u2VZVsjHgtXY2CKfzwp6u4SEGB06dDTo7UwjV3DIFZzemMvugrLj/ezO++V0Os74B7ptlzfHx8crNTVVlZWVkqTKykqlpqYqLi7OyBgAwB4Oy7KC/7M9SCtWrNCWLVt0+PBhDRo0SLGxsXrllVe0Z88eLVmyRN99950GDBigoqIiXXjhhZJkZCwYHNGEB7mCQ67gdPeIZso95SFOFJiKh3N61RFNWIrmbETRhAe5gkOu4FA0wel1U2cAgHMDRQMAMIqiAQAYRdEAAIyiaAAARlE0AACjKBoAgFEUDQDAKIoGAGAURQMAMIqiAQAYRdEAAIyiaAAARlE0AACjKBoAgFEUDQDAKIoGAGAURQMAMIqiAQAYRdEAAIyiaAAARlE0AACjKBoAgFGRdgcAgJ8TM6Cv+kR1/X9VCQkxIUyDrqBoAPRofaIiNeWe8rDvt+LhnLDvs7di6gwAYFSPOKLJzMyUy+VSVFSUJGnhwoUaO3asPvnkE+Xn56utrU3JyclatWqV4uPjJanLYwCA8OoxRzSPP/64ysvLVV5errFjx8rn82nRokXKz89XdXW1MjIytHr1aknq8hgAIPx6TNH8r9raWkVFRSkjI0OSNGPGDL322mvdGgMAhF+PmDqTOqbLLMtSenq6FixYII/Ho6SkJP94XFycfD6fjhw50uWx2NjYsL4mAEAPKZoNGzbI7Xbr5MmTWrlypQoLCzV+/HhbM8XH9+/ytj31ckpyBYdcwempuc5Wdr2fJvbbI4rG7XZLklwul3JzczVnzhzddNNNqq+v9z+mqalJTqdTsbGxcrvdXRoLRmNji3w+K+jXkpAQo0OHjga9nWnkCg65gmMy17laYHb8nLvzc3Q6HWf8A932czTHjx/X0aMdL8yyLFVVVSk1NVWjRo3SiRMntH37dknSxo0bNXHiREnq8hgAIPxsP6JpbGxUXl6evF6vfD6fRo4cqYKCAjmdThUXF6ugoKDTZcqSujwGAAg/24tm+PDhKisr+8mxK664QhUVFSEdAwCEl+1TZwCA3o2iAQAYRdEAAIyiaAAARlE0AACjKBoAgFEUDQDAKIoGAGAURQMAMIqiAQAYZfsSNACAzk5+77Vl1eqT33uNPC9FAwA9jOu8CE25pzzs+614OMfI8zJ1BgAwiqIBABhF0QAAjKJoAABGUTQAAKMoGgCAURQNAMAoigYAYBQf2AQQkJgBfdUn6sz/y7Djk+w4O1A0AALSJyqyV31aHeHD1BkAwCiOaICziF2LLQLdQdEAZxG7FluUmMJC11E0QBf80olxAD/otb8pe/fu1ZIlS3TkyBHFxsaqqKhII0aMsDsWeglOjAOB67VFU1BQoNzcXOXk5Ki8vFz5+fl64YUX7I6FEDJ9VMG5ECA0emXRNDY2qq6uTs8++6wkKTs7W8uXL1dTU5Pi4uICeg6n09Glfdt1sratrV0tLSd+9jFdfU2mdTVXn6hI3bJiS4jTBGb9st8qcVBfW/Zt137t3Pe5tl87993V38ef285hWZbV1UA9VW1tre6991698sor/vuuu+46rVq1SpdccomNyQDg3MPnaAAARvXKonG73Tp48KC8Xq8kyev1qqGhQW632+ZkAHDu6ZVFEx8fr9TUVFVWVkqSKisrlZqaGvD5GQBA6PTKczSStGfPHi1ZskTfffedBgwYoKKiIl144YV2xwKAc06vLRoAQM/QK6fOAAA9B0UDADCKogEAGEXRAACM6pVL0Nihpy7iWVRUpOrqah04cEAVFRW6+OKL7Y6k5uZmLV68WN98841cLpfOP/98FRYW9ojLz+fOnav9+/fL6XQqOjpa999/v1JTU+2O5ffEE09ozZo1PeZnKUmZmZlyuVyKioqSJC1cuFBjx461NVNbW5sefPBBvf/++4qKilJaWpqWL19uayZJ2r9/v+68807/7aNHj6qlpUX/+c9/bEzV4c0339Rjjz0my7JkWZbmzZun3/72t6F5cgshceONN1plZWWWZVlWWVmZdeONN9qcqMO2bdus+vp669prr7W++OILu+NYlmVZzc3N1gcffOC//fe//93661//amOiH3z33Xf+f7/++uvWtGnTbEzTWW1trXXLLbf0qJ+lZVk9Lo9lWdby5cutlStXWj6fz7Isyzp06JDNiX7aihUrrL/97W92x7B8Pp+VkZHh/znu3LnTSktLs7xeb0ien6mzEDi1iGd2drakjkU86+rq1NTUZHMyKSMjo8etiBAbG6vRo0f7b6elpam+vt7GRD+IiflhQdSWlhY5HD1jIdKTJ0+qsLBQDzzwgN1Rerxjx46prKxM8+fP9//8Bg8ebHOqHzt58qQqKip0/fXX2x1FkuR0OnX06FFJHUdaiYmJcjpDUxFMnYWAx+PRkCFDFBERIUmKiIhQYmKiPB5Pj5gO6sl8Pp9efPFFZWZm2h3Fb+nSpdq6dassy9IzzzxjdxxJ0mOPPaapU6dq2LBhdkf5SQsXLpRlWUpPT9eCBQs0YMAA27Ls27dPsbGxeuKJJ/Thhx+qX79+mj9/vjIyMmzL9FNqamo0ZMiQHrHQr8Ph0KOPPqq5c+cqOjpax44d01NPPRWy5+eIBrZavny5oqOjNXPmTLuj+K1cuVJvvfWW7r77bhUXF9sdRx9//LFqa2uVm5trd5SftGHDBm3evFkvvfSSLMtSYWGhrXm8Xq/27dunX/3qV3r55Ze1cOFC5eXlqaWlxdZc/+ull17qMUcz7e3tevLJJ7V27Vq9+eabWrdune666y4dO3YsJM9P0YQAi3h2TVFRkb7++ms9+uijITtED6Vp06bpww8/VHNzs605tm3bpj179igrK0uZmZn69ttvdcstt+jdd9+1Ndcpp/47d7lcys3N1UcffWR7nsjISP9U9mWXXaZBgwZp7969tuY63cGDB7Vt2zZNmTLF7iiSpJ07d6qhoUHp6emSpPT0dPXt21d79uwJyfP3vN/usxCLeAbvkUceUW1trUpKSuRyueyOI6ljbt/j8fhv19TUaODAgYqNjbUxlXTbbbfp3XffVU1NjWpqajR06FCtX79eY8aMsTWXJB0/ftw/r29Zlqqqqmy/Si8uLk6jR4/W1q1bJXVcEdrY2Kjzzz/f1lyn27Rpk8aNG6dBgwbZHUWSNHToUH377bf68ssvJXWsFdnY2Kj/+7//C8nzs9ZZiPTURTxXrFihLVu26PDhwxo0aJBiY2M7fSGcHXbv3q3s7GyNGDFCffr0kSQNGzZMJSUltuY6fPiw5s6dq9bWVjmdTg0cOFD33ntvj5hDP11mZqZKS0t7xOXN+/btU15enrxer3w+n0aOHKlly5YpMTHR9lz33Xefjhw5osjISN11110aN26crZlON2HCBC1dulTXXHON3VH8Nm/erKefftp/AcVf/vIX/eY3vwnJc1M0AACjmDoDABhF0QAAjKJoAABGUTQAAKMoGgCAURQNAMAoigboQTIzM/Xee+/ZHQMIKYoGOIu0t7fbHQEIGkUDGHDw4EHl5eXpqquuUmZmpl544QVJ0po1azR//nwtXrxYl19+uSZPnqxPP/1UkrRo0SLV19frjjvu0OWXX66nn35a+/fvV0pKiv71r3/p17/+tf70pz/J5/Np7dq1uvbaa3X11Vdr8eLF/mVgTj3+n//8p8aMGaMxY8Zo/fr1kqRDhw7psssu67R222effaarrrpK33//fZjfIZxLKBogxHw+n+bMmaOUlBS9/fbbev755/X888/rnXfekdSxhtrkyZO1fft2ZWZm+r/5cdWqVUpKSlJpaak+/vhj3Xrrrf7n3LZtm6qqqrR+/Xq9/PLL2rRpk1544QW98cYbOn78+I9WTP7www+1ZcsWrV+/Xk8//bTee+89JSQk6Morr9Srr77qf1x5ebkmT56s8847LwzvDM5VFA0QYp9++qmampo0b948uVwuDR8+XH/84x9VVVUlqWNl3HHjxikiIkI5OTn6/PPPf/E58/LyFB0drT59+qiiokI333yzhg8frn79+mnBggWqqqrqNK125513Kjo6WikpKfr973/vX/D1d7/7nTZv3iypY5XxV155RTk5OQbeBeAHfPEZEGIHDhxQQ0NDpy/a8nq9ysjIUFJSUqdve+zTp4/a2trU3t6uyMgz/zoOHTrU/++GhgYlJyf7bycnJ6u9vV2NjY3++07/iork5GTt2rVLkpSVlaWCggLt27dPe/fuVf/+/XXppZd27wUDv4CiAULM7XZr2LBh2rJly4/G1qxZ06XnPP0rpRMTE3XgwAH/7fr6ekVGRio+Pl7ffvutpI5vfR05cqR//NRqylFRUZo0aZI2b96sL7/8kqMZhAVTZ0CIXXrpperXr5+eeuopnThxQl6vV7t27dKOHTt+cdvBgwdr3759P/uY7OxsPf/889q3b5+OHTumf/zjH5o0aVKnI6K1a9eqtbVVu3fv1ssvv6zrrrvOP5aTk6NNmzappqaGokFYUDRAiEVERKi0tFSff/65srKydNVVV2nZsmUBfZXwbbfdpnXr1ikjI8N/tdj/uv766zV16lTNnDlTWVlZcrlcuv/++zs95sorr9T48eN18803a9asWZ2+JC09PV1Op1OXXHJJpyk4wBS+jwboRfbv36+srCx99tlnP3vO56abbtKUKVN0ww03hDEdzlUc0QDnmB07dqiurk6TJk2yOwrOEVwMAJxD7r33Xr3xxhtaunSp+vfvb3ccnCOYOgMAGMXUGQDAKIoGAGAURQMAMIqiAQAYRdEAAIyiaAAARv0/OJulEIF9O8YAAAAASUVORK5CYII=\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"UQmEz03oaDWZ","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":296},"outputId":"0f126f42-0dc7-48b6-ea99-9d9144c33e12","executionInfo":{"status":"ok","timestamp":1592059570651,"user_tz":-480,"elapsed":21136,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}}},"source":["# data size distribution\n","from matplotlib.colors import ListedColormap\n","colors = ListedColormap(['steelblue','indianred'])\n","sc = plt.scatter(train_df[\"size_of_data\"], train_df[\"virtual_size\"], c=train_df[\"malware\"], cmap=colors)\n","plt.legend(handles=sc.legend_elements()[0], labels=[\"Benign\", 'Malware'])\n","plt.xlabel(\"size of data\")\n","plt.ylabel(\"virtual size\")\n","plt.show()"],"execution_count":16,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"id":"164TdfYdxfsb","colab_type":"text"},"source":["### ADASYN Resampling"]},{"cell_type":"code","metadata":{"id":"DXUJsE97SZ8m","colab_type":"code","colab":{}},"source":["# Drop hash\n","X = np.array(train_df.iloc[:, 1:-1])\n","y = np.array(train_df.malware)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"VRPlaBK3R9k5","colab_type":"code","outputId":"92f8afb1-c19c-4bce-b9c8-cfb1190bcc4f","executionInfo":{"status":"ok","timestamp":1592059593215,"user_tz":-480,"elapsed":43671,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":139}},"source":["# Do ADASYN over sampling\n","from imblearn.over_sampling import ADASYN\n","ada = ADASYN(random_state=42)\n","X, y = ada.fit_resample(X, y)"],"execution_count":18,"outputs":[{"output_type":"stream","text":["/usr/local/lib/python3.6/dist-packages/sklearn/externals/six.py:31: FutureWarning: The module is deprecated in version 0.21 and will be removed in version 0.23 since we've dropped support for Python 2.7. Please rely on the official version of six (https://pypi.org/project/six/).\n","  \"(https://pypi.org/project/six/).\", FutureWarning)\n","/usr/local/lib/python3.6/dist-packages/sklearn/utils/deprecation.py:144: FutureWarning: The sklearn.neighbors.base module is  deprecated in version 0.22 and will be removed in version 0.24. The corresponding classes / functions should instead be imported from sklearn.neighbors. Anything that cannot be imported from sklearn.neighbors is now part of the private API.\n","  warnings.warn(message, FutureWarning)\n","/usr/local/lib/python3.6/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function safe_indexing is deprecated; safe_indexing is deprecated in version 0.22 and will be removed in version 0.24.\n","  warnings.warn(msg, category=FutureWarning)\n"],"name":"stderr"}]},{"cell_type":"code","metadata":{"id":"lTcZZsZaES_n","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":285},"outputId":"b4132cb5-b243-42ab-db45-cfefa1a5f140","executionInfo":{"status":"ok","timestamp":1592059593216,"user_tz":-480,"elapsed":43658,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}}},"source":["# malware vs. benign distribution after ADASYN resampling\n","plt.bar([0,1], [len(y[y==0]), len(y[y==1])], color=\"brown\")\n","plt.xticks([0,1])\n","plt.xlabel(\"malware (0: benign / 1:malware)\")\n","plt.ylabel(\"count\")\n","plt.show()"],"execution_count":19,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"KZZV01XGE-E9","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":285},"outputId":"8a642de1-1288-48bd-82be-534e25a7aa1c","executionInfo":{"status":"ok","timestamp":1592059593217,"user_tz":-480,"elapsed":43641,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}}},"source":["# entropy distribution after ADASYN resampling\n","plt.hist(X[:, 2], color=\"brown\")\n","plt.xlabel(\"entropy\")\n","plt.ylabel(\"count\")\n","plt.show()"],"execution_count":20,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"yGhhFlzl6wxe","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":296},"outputId":"3f000337-5089-4dab-9a56-bea2784616da","executionInfo":{"status":"ok","timestamp":1592059596726,"user_tz":-480,"elapsed":47135,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}}},"source":["# data size distribution after ADASYN resampling\n","sc = plt.scatter(X[:, 0], X[:, 3], c=y, cmap=colors)\n","plt.legend(handles=sc.legend_elements()[0], labels=[\"Benign\", 'Malware'])\n","plt.xlabel(\"size of data\")\n","plt.ylabel(\"virtual size\")\n","plt.show()"],"execution_count":21,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"id":"LEKvzNgSQly5","colab_type":"text"},"source":["## Part2. Header Section Data -- Build DNN Model"]},{"cell_type":"code","metadata":{"id":"IQ6RykFz-Tx6","colab_type":"code","outputId":"e693c180-e8fd-4f4a-b33f-8642f4a62626","executionInfo":{"status":"ok","timestamp":1592059596728,"user_tz":-480,"elapsed":47122,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":187}},"source":["# Header Data\n","header_X = X[:, 0:5]\n","print(header_X[:5])"],"execution_count":22,"outputs":[{"output_type":"stream","text":["[[4.43392000e+05 4.09600000e+03 6.52962359e+00 4.42984000e+05\n","  1.00092103e+00]\n"," [3.31264000e+05 4.09600000e+03 6.60431385e+00 3.30784000e+05\n","  1.00145110e+00]\n"," [7.42400000e+04 4.09600000e+03 6.04678910e+00 7.38190000e+04\n","  1.00570314e+00]\n"," [2.19648000e+05 4.09600000e+03 6.49701757e+00 2.19524000e+05\n","  1.00056486e+00]\n"," [2.62144000e+05 4.09600000e+03 6.63814158e+00 2.61943000e+05\n","  1.00076734e+00]]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"8RI0ArYlVXQG","colab_type":"code","outputId":"0d60baba-fa86-4fdf-9ec0-7529c76e6cd7","executionInfo":{"status":"ok","timestamp":1592059596730,"user_tz":-480,"elapsed":47108,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":102}},"source":["# Standardization\n","from sklearn.preprocessing import StandardScaler\n","header_X = StandardScaler().fit_transform(header_X)\n","print(header_X[:5])"],"execution_count":23,"outputs":[{"output_type":"stream","text":["[[-0.10569551 -0.03205049  0.23969828 -0.10764297 -0.01898153]\n"," [-0.17650603 -0.03205049  0.34858063 -0.17824818 -0.01756819]\n"," [-0.33882056 -0.03205049 -0.46417062 -0.33995112 -0.00623086]\n"," [-0.24699322 -0.03205049  0.19216571 -0.24826188 -0.01993119]\n"," [-0.22015635 -0.03205049  0.39789419 -0.22156845 -0.0193913 ]]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"zThb6JS1ODQs","colab_type":"code","colab":{}},"source":["def header_model():\n","    model = keras.models.Sequential(name=\"header_dnn\")\n","    model.add(keras.layers.Dense(32, input_dim=header_X.shape[1], activation=\"relu\"))\n","    model.add(keras.layers.Dense(32, activation=\"relu\"))\n","    model.add(keras.layers.Dense(64, activation=\"relu\"))\n","    model.add(keras.layers.Dropout(0.2))\n","    model.add(keras.layers.Dense(1, activation=\"sigmoid\"))\n","    return model"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"I1Za8URkFyS8","colab_type":"code","outputId":"cd3abea3-7a14-4dee-9538-0d883eae9de4","executionInfo":{"status":"ok","timestamp":1592059598563,"user_tz":-480,"elapsed":48914,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":323}},"source":["header_dnn = header_model()\n","header_dnn.summary()"],"execution_count":25,"outputs":[{"output_type":"stream","text":["Model: \"header_dnn\"\n","_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","dense (Dense)                (None, 32)                192       \n","_________________________________________________________________\n","dense_1 (Dense)              (None, 32)                1056      \n","_________________________________________________________________\n","dense_2 (Dense)              (None, 64)                2112      \n","_________________________________________________________________\n","dropout (Dropout)            (None, 64)                0         \n","_________________________________________________________________\n","dense_3 (Dense)              (None, 1)                 65        \n","=================================================================\n","Total params: 3,425\n","Trainable params: 3,425\n","Non-trainable params: 0\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"hRH892vqHV9d","colab_type":"code","colab":{}},"source":["def compile_and_train(model, X, alpha, patience):\n","    model.compile(optimizer=keras.optimizers.Adam(learning_rate=alpha), \n","              loss=keras.losses.BinaryCrossentropy(), \n","              metrics=['accuracy'])\n","    file_path = model.name + \".hdf5\"\n","    early_stopping = keras.callbacks.EarlyStopping(monitor='val_loss', patience=patience)\n","    checkpoint = keras.callbacks.ModelCheckpoint(file_path, monitor='loss', verbose=0, save_weights_only=True, save_best_only=True, mode='auto', period=1)\n","    history = model.fit(X, y, epochs=300, callbacks=[early_stopping, checkpoint], validation_split=0.4, batch_size=80, shuffle=True)\n","    model.save(file_path)\n","    return history"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"IDMufpt2KnKn","colab_type":"code","outputId":"460c8eca-7b05-47e6-e76c-5fdd84587225","executionInfo":{"status":"ok","timestamp":1592059771320,"user_tz":-480,"elapsed":221642,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":1000}},"source":["header_history = compile_and_train(header_dnn, header_X, 3e-4, 20)"],"execution_count":27,"outputs":[{"output_type":"stream","text":["WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n","Epoch 1/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.4301 - accuracy: 0.8350 - val_loss: 1.4302 - val_accuracy: 0.0000e+00\n","Epoch 2/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3577 - accuracy: 0.8413 - val_loss: 1.4033 - val_accuracy: 0.1229\n","Epoch 3/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3512 - accuracy: 0.8477 - val_loss: 1.3663 - val_accuracy: 0.1471\n","Epoch 4/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3469 - accuracy: 0.8486 - val_loss: 1.4373 - val_accuracy: 0.1087\n","Epoch 5/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3433 - accuracy: 0.8492 - val_loss: 1.2795 - val_accuracy: 0.1887\n","Epoch 6/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3396 - accuracy: 0.8498 - val_loss: 1.2906 - val_accuracy: 0.1665\n","Epoch 7/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3357 - accuracy: 0.8511 - val_loss: 1.3106 - val_accuracy: 0.1726\n","Epoch 8/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3325 - accuracy: 0.8524 - val_loss: 1.2536 - val_accuracy: 0.2069\n","Epoch 9/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3290 - accuracy: 0.8525 - val_loss: 1.1830 - val_accuracy: 0.2323\n","Epoch 10/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.3248 - accuracy: 0.8548 - val_loss: 1.1244 - val_accuracy: 0.2760\n","Epoch 11/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3210 - accuracy: 0.8561 - val_loss: 1.2012 - val_accuracy: 0.2379\n","Epoch 12/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3176 - accuracy: 0.8571 - val_loss: 1.2764 - val_accuracy: 0.2272\n","Epoch 13/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3139 - accuracy: 0.8581 - val_loss: 1.2202 - val_accuracy: 0.2462\n","Epoch 14/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3108 - accuracy: 0.8600 - val_loss: 1.1205 - val_accuracy: 0.3042\n","Epoch 15/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3072 - accuracy: 0.8641 - val_loss: 1.0985 - val_accuracy: 0.2962\n","Epoch 16/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3051 - accuracy: 0.8687 - val_loss: 1.1525 - val_accuracy: 0.2922\n","Epoch 17/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.3019 - accuracy: 0.8729 - val_loss: 1.0765 - val_accuracy: 0.5287\n","Epoch 18/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2995 - accuracy: 0.8770 - val_loss: 1.2204 - val_accuracy: 0.4460\n","Epoch 19/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2969 - accuracy: 0.8798 - val_loss: 1.1833 - val_accuracy: 0.4547\n","Epoch 20/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2953 - accuracy: 0.8806 - val_loss: 1.0600 - val_accuracy: 0.5097\n","Epoch 21/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2935 - accuracy: 0.8822 - val_loss: 1.2097 - val_accuracy: 0.4620\n","Epoch 22/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2915 - accuracy: 0.8833 - val_loss: 1.1142 - val_accuracy: 0.4708\n","Epoch 23/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.2899 - accuracy: 0.8841 - val_loss: 1.1926 - val_accuracy: 0.4813\n","Epoch 24/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.2887 - accuracy: 0.8840 - val_loss: 1.1205 - val_accuracy: 0.4878\n","Epoch 25/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.2872 - accuracy: 0.8847 - val_loss: 1.0431 - val_accuracy: 0.5480\n","Epoch 26/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.2864 - accuracy: 0.8851 - val_loss: 1.1698 - val_accuracy: 0.4835\n","Epoch 27/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2854 - accuracy: 0.8849 - val_loss: 1.1203 - val_accuracy: 0.5140\n","Epoch 28/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2837 - accuracy: 0.8861 - val_loss: 1.0649 - val_accuracy: 0.5261\n","Epoch 29/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2825 - accuracy: 0.8864 - val_loss: 1.0802 - val_accuracy: 0.5339\n","Epoch 30/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2812 - accuracy: 0.8876 - val_loss: 1.1294 - val_accuracy: 0.5008\n","Epoch 31/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2810 - accuracy: 0.8865 - val_loss: 1.0802 - val_accuracy: 0.4983\n","Epoch 32/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2796 - accuracy: 0.8861 - val_loss: 1.1506 - val_accuracy: 0.5106\n","Epoch 33/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2800 - accuracy: 0.8869 - val_loss: 1.0399 - val_accuracy: 0.5552\n","Epoch 34/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2785 - accuracy: 0.8873 - val_loss: 1.0208 - val_accuracy: 0.5577\n","Epoch 35/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2778 - accuracy: 0.8877 - val_loss: 1.1269 - val_accuracy: 0.4942\n","Epoch 36/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2769 - accuracy: 0.8884 - val_loss: 1.0784 - val_accuracy: 0.5123\n","Epoch 37/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2754 - accuracy: 0.8883 - val_loss: 1.0028 - val_accuracy: 0.5508\n","Epoch 38/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2754 - accuracy: 0.8882 - val_loss: 1.1386 - val_accuracy: 0.4758\n","Epoch 39/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2739 - accuracy: 0.8892 - val_loss: 1.2315 - val_accuracy: 0.4588\n","Epoch 40/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2737 - accuracy: 0.8881 - val_loss: 1.0223 - val_accuracy: 0.5807\n","Epoch 41/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2734 - accuracy: 0.8894 - val_loss: 0.9645 - val_accuracy: 0.6047\n","Epoch 42/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2716 - accuracy: 0.8894 - val_loss: 1.1634 - val_accuracy: 0.4785\n","Epoch 43/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2712 - accuracy: 0.8897 - val_loss: 0.9457 - val_accuracy: 0.5894\n","Epoch 44/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2706 - accuracy: 0.8895 - val_loss: 1.1145 - val_accuracy: 0.4883\n","Epoch 45/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2702 - accuracy: 0.8898 - val_loss: 1.0813 - val_accuracy: 0.5751\n","Epoch 46/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2690 - accuracy: 0.8902 - val_loss: 1.0613 - val_accuracy: 0.5294\n","Epoch 47/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2682 - accuracy: 0.8919 - val_loss: 0.9686 - val_accuracy: 0.5798\n","Epoch 48/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2681 - accuracy: 0.8913 - val_loss: 0.9648 - val_accuracy: 0.5674\n","Epoch 49/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2678 - accuracy: 0.8918 - val_loss: 1.0238 - val_accuracy: 0.5235\n","Epoch 50/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2665 - accuracy: 0.8917 - val_loss: 0.9721 - val_accuracy: 0.5799\n","Epoch 51/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2664 - accuracy: 0.8912 - val_loss: 0.9810 - val_accuracy: 0.5629\n","Epoch 52/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2653 - accuracy: 0.8914 - val_loss: 1.1235 - val_accuracy: 0.4887\n","Epoch 53/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2653 - accuracy: 0.8924 - val_loss: 1.0347 - val_accuracy: 0.5325\n","Epoch 54/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2643 - accuracy: 0.8920 - val_loss: 0.9556 - val_accuracy: 0.6006\n","Epoch 55/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2646 - accuracy: 0.8917 - val_loss: 1.0957 - val_accuracy: 0.5003\n","Epoch 56/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2638 - accuracy: 0.8930 - val_loss: 0.9422 - val_accuracy: 0.5955\n","Epoch 57/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2632 - accuracy: 0.8921 - val_loss: 1.0205 - val_accuracy: 0.5674\n","Epoch 58/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2629 - accuracy: 0.8927 - val_loss: 0.9248 - val_accuracy: 0.5870\n","Epoch 59/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2622 - accuracy: 0.8928 - val_loss: 1.0208 - val_accuracy: 0.5651\n","Epoch 60/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2607 - accuracy: 0.8938 - val_loss: 0.9494 - val_accuracy: 0.5897\n","Epoch 61/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2608 - accuracy: 0.8939 - val_loss: 0.9835 - val_accuracy: 0.5795\n","Epoch 62/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2601 - accuracy: 0.8936 - val_loss: 1.0324 - val_accuracy: 0.5479\n","Epoch 63/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.2599 - accuracy: 0.8943 - val_loss: 1.0215 - val_accuracy: 0.5464\n","Epoch 64/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.2592 - accuracy: 0.8937 - val_loss: 0.8979 - val_accuracy: 0.6083\n","Epoch 65/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2589 - accuracy: 0.8942 - val_loss: 0.9564 - val_accuracy: 0.5670\n","Epoch 66/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2577 - accuracy: 0.8948 - val_loss: 1.0326 - val_accuracy: 0.5567\n","Epoch 67/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2579 - accuracy: 0.8948 - val_loss: 1.0469 - val_accuracy: 0.5443\n","Epoch 68/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2572 - accuracy: 0.8951 - val_loss: 0.9122 - val_accuracy: 0.5879\n","Epoch 69/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2565 - accuracy: 0.8958 - val_loss: 1.0242 - val_accuracy: 0.5465\n","Epoch 70/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2568 - accuracy: 0.8949 - val_loss: 0.9602 - val_accuracy: 0.5694\n","Epoch 71/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2560 - accuracy: 0.8951 - val_loss: 0.9937 - val_accuracy: 0.5325\n","Epoch 72/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2551 - accuracy: 0.8943 - val_loss: 0.9508 - val_accuracy: 0.5792\n","Epoch 73/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2559 - accuracy: 0.8952 - val_loss: 1.0110 - val_accuracy: 0.5377\n","Epoch 74/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2547 - accuracy: 0.8955 - val_loss: 0.9915 - val_accuracy: 0.5714\n","Epoch 75/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2534 - accuracy: 0.8961 - val_loss: 0.9858 - val_accuracy: 0.5761\n","Epoch 76/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2530 - accuracy: 0.8968 - val_loss: 0.9689 - val_accuracy: 0.5502\n","Epoch 77/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2527 - accuracy: 0.8960 - val_loss: 0.9456 - val_accuracy: 0.5791\n","Epoch 78/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2530 - accuracy: 0.8965 - val_loss: 1.0427 - val_accuracy: 0.5358\n","Epoch 79/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2534 - accuracy: 0.8964 - val_loss: 0.9455 - val_accuracy: 0.5839\n","Epoch 80/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2512 - accuracy: 0.8963 - val_loss: 0.9417 - val_accuracy: 0.5733\n","Epoch 81/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2516 - accuracy: 0.8965 - val_loss: 1.0595 - val_accuracy: 0.5489\n","Epoch 82/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2512 - accuracy: 0.8961 - val_loss: 0.9351 - val_accuracy: 0.6036\n","Epoch 83/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2513 - accuracy: 0.8962 - val_loss: 1.0221 - val_accuracy: 0.5745\n","Epoch 84/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.2502 - accuracy: 0.8971 - val_loss: 0.9328 - val_accuracy: 0.5846\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"1X9d7bdjPrJp","colab_type":"code","outputId":"a8e95b95-454f-41a4-a368-5220cd18f242","executionInfo":{"status":"ok","timestamp":1592059771321,"user_tz":-480,"elapsed":221625,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":302}},"source":["# The change of training accuracy and validation accuracy\n","print(\"The training accuracy of section header model: {}%\".format(round(header_history.history[\"accuracy\"][-1]*100, 2)))\n","print(\"The validation accuracy of section header model: {}%\".format(round(header_history.history[\"val_accuracy\"][-1]*100, 2)))\n","\n","\n","sns.set()\n","\n","header_acc = header_history.history[\"accuracy\"]\n","header_val_acc = header_history.history[\"val_accuracy\"]\n","\n","plt.plot(header_acc, \"b-\", label=\"acc\")\n","plt.plot(header_val_acc, \"r-\", label=\"val_acc\")\n","plt.legend()\n","plt.show()"],"execution_count":28,"outputs":[{"output_type":"stream","text":["The training accuracy of section header model: 89.71%\n","The validation accuracy of section header model: 58.46%\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"id":"eP9R7kRUa33u","colab_type":"text"},"source":["## Part3. Import Data -- Build DNN Model"]},{"cell_type":"code","metadata":{"id":"n83xZOOsbDE1","colab_type":"code","outputId":"ffc2b9e4-c5cf-4c4f-e676-9141ed7cea5b","executionInfo":{"status":"ok","timestamp":1592059771322,"user_tz":-480,"elapsed":221609,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":102}},"source":["# Import Data\n","import_X = X[:, 5:1005]\n","print(import_X[:5])"],"execution_count":29,"outputs":[{"output_type":"stream","text":["[[1. 1. 1. ... 0. 0. 0.]\n"," [1. 1. 1. ... 0. 0. 0.]\n"," [1. 1. 1. ... 0. 0. 0.]\n"," [1. 1. 1. ... 0. 0. 0.]\n"," [1. 1. 1. ... 0. 0. 0.]]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"1M6RrwkhW9gc","colab_type":"code","outputId":"5f429b13-4e16-4c59-92ca-353b41652e32","executionInfo":{"status":"ok","timestamp":1592059792106,"user_tz":-480,"elapsed":242375,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":170}},"source":["# Use PCA on import data\n","from sklearn.decomposition import PCA\n","from sklearn.preprocessing import StandardScaler\n","\n","pca = PCA()\n","pca.fit_transform(import_X)\n","evr = pca.explained_variance_ratio_\n","\n","print(evr[:50])"],"execution_count":30,"outputs":[{"output_type":"stream","text":["[0.30551161 0.14745809 0.12729677 0.07992873 0.03218743 0.03005649\n"," 0.02438346 0.01386609 0.0125088  0.01095256 0.00832843 0.00637474\n"," 0.00576745 0.00516138 0.00491347 0.00453915 0.00431081 0.00413218\n"," 0.0038529  0.00377605 0.0033852  0.00323161 0.00299237 0.00292605\n"," 0.00282783 0.00257709 0.00249517 0.00237953 0.00232765 0.00223498\n"," 0.00213049 0.00207172 0.00201726 0.00189608 0.00182543 0.00169457\n"," 0.00167864 0.00161015 0.00156334 0.00154187 0.00144195 0.00139526\n"," 0.00136178 0.00134466 0.00131855 0.00129518 0.00128262 0.00125361\n"," 0.00120877 0.00117217]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"dcDH2xVXYeIi","colab_type":"code","outputId":"26c23606-0e59-4c67-dbda-d57405154b32","executionInfo":{"status":"ok","timestamp":1592059792107,"user_tz":-480,"elapsed":242342,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":285}},"source":["sns.set()\n","def plotData(data):\n","  x = [p[0] for p in data]\n","  y = [p[1] for p in data]\n","  plt.plot(x, y, '-o')\n","\n","plotData(list(zip(range(1, 30), evr)))\n","plt.xlabel(\"PC\")\n","plt.ylabel(\"Explained Variance Ratio\")\n","plt.show()"],"execution_count":31,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"4H4uP3DOavLd","colab_type":"code","outputId":"e7358680-7aeb-4989-e550-91fce89f835d","executionInfo":{"status":"ok","timestamp":1592059792107,"user_tz":-480,"elapsed":242317,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":285}},"source":["# 主成份的累計貢獻率\n","tev = [0]*len(evr)\n","for i in range(len(evr)):\n","    if i == 0:\n","        tev[i] = evr[i]\n","    else:\n","        tev[i] = tev[i-1] + evr[i]\n","plt.plot(tev, \"-o\", c=\"brown\")\n","plt.xlim(0, 50)\n","plt.xlabel(\"Number of components\")\n","plt.ylabel(\"Total explained variance\")\n","plt.yticks(np.arange(0, 1, .1))\n","plt.show()"],"execution_count":32,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"code","metadata":{"id":"OTt-1Z-_cae-","colab_type":"code","outputId":"81a30593-64b8-43c9-87a4-c85cfd19a98c","executionInfo":{"status":"ok","timestamp":1592059792108,"user_tz":-480,"elapsed":242300,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":170}},"source":["def print_pca_result(n):\n","    total_percent = sum(evr[:n])*100\n","    rounded_total_percent = round(total_percent, 2)\n","    print(\"選擇 {} 個主成份，可以解釋的總變異為 {} %\".format(n, rounded_total_percent))\n","\n","for i in range(10, 100, 10):\n","    print_pca_result(i)"],"execution_count":33,"outputs":[{"output_type":"stream","text":["選擇 10 個主成份，可以解釋的總變異為 78.42 %\n","選擇 20 個主成份，可以解釋的總變異為 83.53 %\n","選擇 30 個主成份，可以解釋的總變異為 86.27 %\n","選擇 40 個主成份，可以解釋的總變異為 88.07 %\n","選擇 50 個主成份，可以解釋的總變異為 89.38 %\n","選擇 60 個主成份，可以解釋的總變異為 90.4 %\n","選擇 70 個主成份，可以解釋的總變異為 91.24 %\n","選擇 80 個主成份，可以解釋的總變異為 91.95 %\n","選擇 90 個主成份，可以解釋的總變異為 92.56 %\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"I1WDsgKdep1-","colab_type":"code","outputId":"d7d0d53b-8291-4016-c68c-3bb3e4aee850","executionInfo":{"status":"ok","timestamp":1592059801930,"user_tz":-480,"elapsed":252097,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":272}},"source":["# 決定選擇 60 個主成份 (PC)\n","import_X = PCA(n_components=60).fit_transform(import_X)\n","print(import_X[0])"],"execution_count":34,"outputs":[{"output_type":"stream","text":["[ 1.03022858e+01 -7.71122709e+00 -4.13981043e+00 -8.28920478e+00\n"," -3.62172906e-01 -2.03734534e-01 -2.63738672e+00 -6.76248040e-01\n"," -1.60456662e-01 -1.74180702e-01 -2.70219828e-01  4.54082300e-02\n","  5.63604032e-01  2.41026210e-01  4.13052781e-01 -9.32879294e-02\n"," -1.18722097e+00 -5.13321025e-04 -6.83138898e-01 -7.43635879e-02\n"," -2.07433925e-02 -1.65012814e-01 -2.90700501e-01 -1.00147704e-01\n"," -5.53896516e-02  3.64944597e-02 -2.55438845e-01  3.76758718e-02\n"," -1.72534462e-03 -1.10433167e-01  1.59273107e-02  9.18042279e-02\n"," -1.44042419e-01  4.55634905e-03  6.93545779e-02 -4.48819248e-02\n","  1.38141855e-02  1.80835089e-03 -7.51645288e-02  1.00270117e-01\n"," -7.80239624e-03  5.88967310e-02  4.51466729e-03 -4.26066774e-02\n","  4.39664258e-03  5.92215105e-02  5.32550158e-03  6.83846197e-02\n"," -1.63216946e-02 -1.47366919e-02 -4.72642497e-02  2.03074975e-02\n","  4.68027190e-02  1.97316503e-02 -2.40711017e-02  1.78861860e-02\n"," -2.81400359e-03 -2.66538053e-02  4.87029034e-02 -1.39985419e-02]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"IzH_Bpi_evet","colab_type":"code","colab":{}},"source":["def import_model():\n","    model = keras.models.Sequential(name=\"import_dnn\")\n","    model.add(keras.layers.Dense(64, input_dim=import_X.shape[1], activation=\"relu\"))\n","    model.add(keras.layers.Dense(64, activation=\"relu\"))\n","    model.add(keras.layers.Dropout(0.4))\n","    model.add(keras.layers.Dense(32, activation=\"relu\"))\n","    model.add(keras.layers.Dense(32, activation=\"relu\"))\n","    model.add(keras.layers.Dropout(0.2))\n","    model.add(keras.layers.Dense(1, activation=\"sigmoid\"))\n","    return model"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"I6-YumoifCAd","colab_type":"code","outputId":"45357b3a-d44e-4839-bf8f-8f904c831197","executionInfo":{"status":"ok","timestamp":1592059801931,"user_tz":-480,"elapsed":252058,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":391}},"source":["import_dnn = import_model()\n","import_dnn.summary()"],"execution_count":36,"outputs":[{"output_type":"stream","text":["Model: \"import_dnn\"\n","_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","dense_4 (Dense)              (None, 64)                3904      \n","_________________________________________________________________\n","dense_5 (Dense)              (None, 64)                4160      \n","_________________________________________________________________\n","dropout_1 (Dropout)          (None, 64)                0         \n","_________________________________________________________________\n","dense_6 (Dense)              (None, 32)                2080      \n","_________________________________________________________________\n","dense_7 (Dense)              (None, 32)                1056      \n","_________________________________________________________________\n","dropout_2 (Dropout)          (None, 32)                0         \n","_________________________________________________________________\n","dense_8 (Dense)              (None, 1)                 33        \n","=================================================================\n","Total params: 11,233\n","Trainable params: 11,233\n","Non-trainable params: 0\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"cRQRAfv8fS9E","colab_type":"code","outputId":"91f86c0d-7783-4a94-eb63-61d810d09597","executionInfo":{"status":"ok","timestamp":1592059921420,"user_tz":-480,"elapsed":371522,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":1000}},"source":["import_history = compile_and_train(import_dnn, import_X, 1e-4, 12)"],"execution_count":37,"outputs":[{"output_type":"stream","text":["WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n","Epoch 1/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.3652 - accuracy: 0.8259 - val_loss: 0.7878 - val_accuracy: 0.6656\n","Epoch 2/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.1915 - accuracy: 0.9257 - val_loss: 0.5476 - val_accuracy: 0.7829\n","Epoch 3/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.1507 - accuracy: 0.9444 - val_loss: 0.4571 - val_accuracy: 0.8542\n","Epoch 4/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.1323 - accuracy: 0.9524 - val_loss: 0.4065 - val_accuracy: 0.8757\n","Epoch 5/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.1219 - accuracy: 0.9578 - val_loss: 0.3959 - val_accuracy: 0.8841\n","Epoch 6/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.1139 - accuracy: 0.9604 - val_loss: 0.3640 - val_accuracy: 0.8945\n","Epoch 7/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.1083 - accuracy: 0.9625 - val_loss: 0.3461 - val_accuracy: 0.9012\n","Epoch 8/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.1043 - accuracy: 0.9645 - val_loss: 0.3698 - val_accuracy: 0.8983\n","Epoch 9/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.0992 - accuracy: 0.9663 - val_loss: 0.3420 - val_accuracy: 0.9073\n","Epoch 10/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0959 - accuracy: 0.9676 - val_loss: 0.3348 - val_accuracy: 0.9099\n","Epoch 11/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0931 - accuracy: 0.9682 - val_loss: 0.3404 - val_accuracy: 0.9114\n","Epoch 12/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0904 - accuracy: 0.9690 - val_loss: 0.3126 - val_accuracy: 0.9176\n","Epoch 13/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0884 - accuracy: 0.9700 - val_loss: 0.3049 - val_accuracy: 0.9206\n","Epoch 14/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0860 - accuracy: 0.9705 - val_loss: 0.3152 - val_accuracy: 0.9181\n","Epoch 15/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0839 - accuracy: 0.9713 - val_loss: 0.2951 - val_accuracy: 0.9247\n","Epoch 16/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0833 - accuracy: 0.9716 - val_loss: 0.3063 - val_accuracy: 0.9207\n","Epoch 17/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0819 - accuracy: 0.9718 - val_loss: 0.2846 - val_accuracy: 0.9270\n","Epoch 18/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0797 - accuracy: 0.9726 - val_loss: 0.3174 - val_accuracy: 0.9221\n","Epoch 19/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0790 - accuracy: 0.9731 - val_loss: 0.2825 - val_accuracy: 0.9284\n","Epoch 20/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0772 - accuracy: 0.9735 - val_loss: 0.3056 - val_accuracy: 0.9255\n","Epoch 21/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0766 - accuracy: 0.9733 - val_loss: 0.2788 - val_accuracy: 0.9326\n","Epoch 22/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0756 - accuracy: 0.9740 - val_loss: 0.2846 - val_accuracy: 0.9313\n","Epoch 23/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.0742 - accuracy: 0.9741 - val_loss: 0.2946 - val_accuracy: 0.9291\n","Epoch 24/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0732 - accuracy: 0.9747 - val_loss: 0.2930 - val_accuracy: 0.9290\n","Epoch 25/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0732 - accuracy: 0.9748 - val_loss: 0.2859 - val_accuracy: 0.9304\n","Epoch 26/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0727 - accuracy: 0.9750 - val_loss: 0.2780 - val_accuracy: 0.9332\n","Epoch 27/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0718 - accuracy: 0.9750 - val_loss: 0.2683 - val_accuracy: 0.9343\n","Epoch 28/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0716 - accuracy: 0.9749 - val_loss: 0.2742 - val_accuracy: 0.9329\n","Epoch 29/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0706 - accuracy: 0.9758 - val_loss: 0.2845 - val_accuracy: 0.9407\n","Epoch 30/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0696 - accuracy: 0.9756 - val_loss: 0.2495 - val_accuracy: 0.9378\n","Epoch 31/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0689 - accuracy: 0.9759 - val_loss: 0.2729 - val_accuracy: 0.9324\n","Epoch 32/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0678 - accuracy: 0.9762 - val_loss: 0.2454 - val_accuracy: 0.9463\n","Epoch 33/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0680 - accuracy: 0.9762 - val_loss: 0.2669 - val_accuracy: 0.9358\n","Epoch 34/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0676 - accuracy: 0.9766 - val_loss: 0.2629 - val_accuracy: 0.9427\n","Epoch 35/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0672 - accuracy: 0.9764 - val_loss: 0.2648 - val_accuracy: 0.9377\n","Epoch 36/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0665 - accuracy: 0.9772 - val_loss: 0.2494 - val_accuracy: 0.9453\n","Epoch 37/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0663 - accuracy: 0.9773 - val_loss: 0.2402 - val_accuracy: 0.9475\n","Epoch 38/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0656 - accuracy: 0.9772 - val_loss: 0.2610 - val_accuracy: 0.9441\n","Epoch 39/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0641 - accuracy: 0.9774 - val_loss: 0.2459 - val_accuracy: 0.9471\n","Epoch 40/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0640 - accuracy: 0.9778 - val_loss: 0.2375 - val_accuracy: 0.9480\n","Epoch 41/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0655 - accuracy: 0.9772 - val_loss: 0.2351 - val_accuracy: 0.9423\n","Epoch 42/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0633 - accuracy: 0.9776 - val_loss: 0.2201 - val_accuracy: 0.9525\n","Epoch 43/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0628 - accuracy: 0.9782 - val_loss: 0.2540 - val_accuracy: 0.9391\n","Epoch 44/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0629 - accuracy: 0.9781 - val_loss: 0.2361 - val_accuracy: 0.9482\n","Epoch 45/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0622 - accuracy: 0.9785 - val_loss: 0.2477 - val_accuracy: 0.9471\n","Epoch 46/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0617 - accuracy: 0.9782 - val_loss: 0.2350 - val_accuracy: 0.9493\n","Epoch 47/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.0618 - accuracy: 0.9781 - val_loss: 0.2349 - val_accuracy: 0.9493\n","Epoch 48/300\n","612/612 [==============================] - 2s 3ms/step - loss: 0.0610 - accuracy: 0.9786 - val_loss: 0.2339 - val_accuracy: 0.9493\n","Epoch 49/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0611 - accuracy: 0.9786 - val_loss: 0.2227 - val_accuracy: 0.9519\n","Epoch 50/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0609 - accuracy: 0.9783 - val_loss: 0.2260 - val_accuracy: 0.9514\n","Epoch 51/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0595 - accuracy: 0.9796 - val_loss: 0.2394 - val_accuracy: 0.9499\n","Epoch 52/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0593 - accuracy: 0.9791 - val_loss: 0.2383 - val_accuracy: 0.9481\n","Epoch 53/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0603 - accuracy: 0.9788 - val_loss: 0.2326 - val_accuracy: 0.9485\n","Epoch 54/300\n","612/612 [==============================] - 2s 4ms/step - loss: 0.0590 - accuracy: 0.9797 - val_loss: 0.2378 - val_accuracy: 0.9426\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"ekFIUf7VgKMw","colab_type":"code","outputId":"df3dfdcb-bc8b-4e05-dfe5-d7182ebc5d91","executionInfo":{"status":"ok","timestamp":1592059921421,"user_tz":-480,"elapsed":371503,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":302}},"source":["# The change of training accuracy and validation accuracy\n","print(\"The training accuracy of section header model: {}%\".format(round(import_history.history[\"accuracy\"][-1]*100, 2)))\n","print(\"The validation accuracy of section header model: {}%\".format(round(import_history.history[\"val_accuracy\"][-1]*100, 2)))\n","\n","sns.set()\n","\n","import_acc = import_history.history[\"accuracy\"]\n","import_val_acc = import_history.history[\"val_accuracy\"]\n","\n","plt.plot(import_acc, \"b-\", label=\"acc\")\n","plt.plot(import_val_acc, \"r-\", label=\"val_acc\")\n","plt.legend()\n","plt.show()"],"execution_count":38,"outputs":[{"output_type":"stream","text":["The training accuracy of section header model: 97.97%\n","The validation accuracy of section header model: 94.26%\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"id":"xLHv5I7ci9c5","colab_type":"text"},"source":["## Part4. Image Data -- Build CNN Model"]},{"cell_type":"code","metadata":{"id":"3UiqD-14jBqS","colab_type":"code","outputId":"0740e940-4933-4086-c069-d6b441348a59","executionInfo":{"status":"ok","timestamp":1592059921422,"user_tz":-480,"elapsed":371482,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":102}},"source":["# Image Data\n","image_X = X[:, 1005:]\n","print(image_X[:5])"],"execution_count":39,"outputs":[{"output_type":"stream","text":["[[137. 137.  16. ... 187. 121.  61.]\n"," [  0.  36.   0. ...   0.   0.   0.]\n"," [192.  69.  12. ...  52.  53.  54.]\n"," [ 35. 141. 232. ...   0.   0.   0.]\n"," [199. 254. 198. ...  61.  63.  50.]]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"vNfRgReD2mQA","colab_type":"code","outputId":"4afd7898-7126-4c54-98b1-45db166a861f","executionInfo":{"status":"ok","timestamp":1592059921423,"user_tz":-480,"elapsed":371466,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":850}},"source":["# Normalization (Min-Max) (From [0,255] to [0, 1])\n","image_X = image_X.astype(\"float32\") / 255.0\n","# Reshape\n","image_X = image_X.reshape(image_X.shape[0], 32, 32, 1)\n","print(image_X[0])"],"execution_count":40,"outputs":[{"output_type":"stream","text":["[[[0.5372549 ]\n","  [0.5372549 ]\n","  [0.0627451 ]\n","  ...\n","  [1.        ]\n","  [0.05882353]\n","  [1.        ]]\n","\n"," [[0.9882353 ]\n","  [0.95686275]\n","  [0.27058825]\n","  ...\n","  [0.8901961 ]\n","  [0.9137255 ]\n","  [0.16862746]]\n","\n"," [[0.        ]\n","  [0.        ]\n","  [0.39607844]\n","  ...\n","  [0.        ]\n","  [0.3137255 ]\n","  [0.        ]]\n","\n"," ...\n","\n"," [[0.22352941]\n","  [0.80784315]\n","  [0.31764707]\n","  ...\n","  [0.8156863 ]\n","  [0.04313726]\n","  [0.84705883]]\n","\n"," [[0.8901961 ]\n","  [0.02745098]\n","  [0.3019608 ]\n","  ...\n","  [0.9372549 ]\n","  [0.3372549 ]\n","  [0.20392157]]\n","\n"," [[0.48235294]\n","  [0.77254903]\n","  [0.49019608]\n","  ...\n","  [0.73333335]\n","  [0.4745098 ]\n","  [0.23921569]]]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"lFyzFbiFkRsq","colab_type":"code","colab":{}},"source":["def image_model():\n","    model = keras.models.Sequential(name=\"image_cnn\")\n","    model.add(keras.layers.Input(shape=(32, 32, 1)))\n","    model.add(keras.layers.Conv2D(32, (4,4), activation=\"relu\", padding=\"same\"))\n","    model.add(keras.layers.Conv2D(64, (4,4), activation=\"relu\", padding=\"same\"))\n","    model.add(keras.layers.MaxPooling2D(2,2))\n","    model.add(keras.layers.Conv2D(128, (4,4), activation=\"relu\", padding=\"same\"))\n","    model.add(keras.layers.Conv2D(128, (4,4), activation=\"relu\", padding=\"same\"))\n","    model.add(keras.layers.MaxPooling2D(2,2))\n","    model.add(keras.layers.Flatten())\n","    model.add(keras.layers.Dense(256, activation=\"relu\"))\n","    model.add(keras.layers.Dropout(0.4))\n","    model.add(keras.layers.Dense(1, activation=\"sigmoid\"))\n","    return model"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"Wu8Sj_IWnhWI","colab_type":"code","colab":{}},"source":["image_cnn = image_model()"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"l6Bm2uIlno1g","colab_type":"code","outputId":"c52cd7ad-4234-4cdf-fa6b-7e0a4655a4d4","executionInfo":{"status":"ok","timestamp":1592059921426,"user_tz":-480,"elapsed":371407,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":493}},"source":["image_cnn.summary()"],"execution_count":43,"outputs":[{"output_type":"stream","text":["Model: \"image_cnn\"\n","_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","conv2d (Conv2D)              (None, 32, 32, 32)        544       \n","_________________________________________________________________\n","conv2d_1 (Conv2D)            (None, 32, 32, 64)        32832     \n","_________________________________________________________________\n","max_pooling2d (MaxPooling2D) (None, 16, 16, 64)        0         \n","_________________________________________________________________\n","conv2d_2 (Conv2D)            (None, 16, 16, 128)       131200    \n","_________________________________________________________________\n","conv2d_3 (Conv2D)            (None, 16, 16, 128)       262272    \n","_________________________________________________________________\n","max_pooling2d_1 (MaxPooling2 (None, 8, 8, 128)         0         \n","_________________________________________________________________\n","flatten (Flatten)            (None, 8192)              0         \n","_________________________________________________________________\n","dense_9 (Dense)              (None, 256)               2097408   \n","_________________________________________________________________\n","dropout_3 (Dropout)          (None, 256)               0         \n","_________________________________________________________________\n","dense_10 (Dense)             (None, 1)                 257       \n","=================================================================\n","Total params: 2,524,513\n","Trainable params: 2,524,513\n","Non-trainable params: 0\n","_________________________________________________________________\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"nsU_UHW2nsqs","colab_type":"code","outputId":"894e55d1-5fa0-45e4-b0d8-046f3133748f","executionInfo":{"status":"ok","timestamp":1592061327149,"user_tz":-480,"elapsed":1777114,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":1000}},"source":["image_history = compile_and_train(image_cnn, image_X, 3e-6, 25)"],"execution_count":44,"outputs":[{"output_type":"stream","text":["WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n","Epoch 1/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.4552 - accuracy: 0.8355 - val_loss: 1.6410 - val_accuracy: 0.0000e+00\n","Epoch 2/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.4008 - accuracy: 0.8357 - val_loss: 1.4966 - val_accuracy: 0.0000e+00\n","Epoch 3/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.3721 - accuracy: 0.8365 - val_loss: 1.4220 - val_accuracy: 0.0098\n","Epoch 4/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.3568 - accuracy: 0.8419 - val_loss: 1.3974 - val_accuracy: 0.0500\n","Epoch 5/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.3453 - accuracy: 0.8484 - val_loss: 1.3334 - val_accuracy: 0.0946\n","Epoch 6/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.3338 - accuracy: 0.8540 - val_loss: 1.2398 - val_accuracy: 0.1659\n","Epoch 7/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.3241 - accuracy: 0.8590 - val_loss: 1.2348 - val_accuracy: 0.1974\n","Epoch 8/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.3151 - accuracy: 0.8650 - val_loss: 1.2248 - val_accuracy: 0.2247\n","Epoch 9/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.3059 - accuracy: 0.8700 - val_loss: 1.1957 - val_accuracy: 0.2546\n","Epoch 10/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.2968 - accuracy: 0.8757 - val_loss: 1.0467 - val_accuracy: 0.3621\n","Epoch 11/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2874 - accuracy: 0.8787 - val_loss: 1.0417 - val_accuracy: 0.3794\n","Epoch 12/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2775 - accuracy: 0.8831 - val_loss: 0.9435 - val_accuracy: 0.4607\n","Epoch 13/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2677 - accuracy: 0.8885 - val_loss: 1.0033 - val_accuracy: 0.4285\n","Epoch 14/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2587 - accuracy: 0.8937 - val_loss: 0.9073 - val_accuracy: 0.5099\n","Epoch 15/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2504 - accuracy: 0.8991 - val_loss: 0.8832 - val_accuracy: 0.5357\n","Epoch 16/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2423 - accuracy: 0.9034 - val_loss: 0.8833 - val_accuracy: 0.5486\n","Epoch 17/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2360 - accuracy: 0.9066 - val_loss: 0.7561 - val_accuracy: 0.6370\n","Epoch 18/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2314 - accuracy: 0.9095 - val_loss: 0.7277 - val_accuracy: 0.6579\n","Epoch 19/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.2266 - accuracy: 0.9118 - val_loss: 0.7466 - val_accuracy: 0.6511\n","Epoch 20/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2224 - accuracy: 0.9145 - val_loss: 0.7655 - val_accuracy: 0.6436\n","Epoch 21/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2189 - accuracy: 0.9162 - val_loss: 0.6155 - val_accuracy: 0.7358\n","Epoch 22/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2159 - accuracy: 0.9176 - val_loss: 0.5920 - val_accuracy: 0.7496\n","Epoch 23/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2130 - accuracy: 0.9194 - val_loss: 0.5533 - val_accuracy: 0.7730\n","Epoch 24/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2100 - accuracy: 0.9202 - val_loss: 0.5543 - val_accuracy: 0.7744\n","Epoch 25/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2072 - accuracy: 0.9221 - val_loss: 0.5934 - val_accuracy: 0.7530\n","Epoch 26/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2058 - accuracy: 0.9230 - val_loss: 0.5670 - val_accuracy: 0.7674\n","Epoch 27/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2028 - accuracy: 0.9247 - val_loss: 0.6165 - val_accuracy: 0.7423\n","Epoch 28/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.2018 - accuracy: 0.9247 - val_loss: 0.5416 - val_accuracy: 0.7821\n","Epoch 29/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1991 - accuracy: 0.9261 - val_loss: 0.5347 - val_accuracy: 0.7844\n","Epoch 30/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1968 - accuracy: 0.9264 - val_loss: 0.5165 - val_accuracy: 0.7932\n","Epoch 31/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1952 - accuracy: 0.9270 - val_loss: 0.4835 - val_accuracy: 0.8132\n","Epoch 32/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1933 - accuracy: 0.9283 - val_loss: 0.5306 - val_accuracy: 0.7863\n","Epoch 33/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1918 - accuracy: 0.9279 - val_loss: 0.6004 - val_accuracy: 0.7551\n","Epoch 34/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1903 - accuracy: 0.9292 - val_loss: 0.4745 - val_accuracy: 0.8176\n","Epoch 35/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1884 - accuracy: 0.9301 - val_loss: 0.4778 - val_accuracy: 0.8155\n","Epoch 36/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1869 - accuracy: 0.9304 - val_loss: 0.5642 - val_accuracy: 0.7697\n","Epoch 37/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1859 - accuracy: 0.9311 - val_loss: 0.3593 - val_accuracy: 0.8762\n","Epoch 38/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1838 - accuracy: 0.9308 - val_loss: 0.5772 - val_accuracy: 0.7646\n","Epoch 39/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1823 - accuracy: 0.9325 - val_loss: 0.5747 - val_accuracy: 0.7651\n","Epoch 40/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1824 - accuracy: 0.9324 - val_loss: 0.4966 - val_accuracy: 0.8046\n","Epoch 41/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1798 - accuracy: 0.9330 - val_loss: 0.5288 - val_accuracy: 0.7874\n","Epoch 42/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1785 - accuracy: 0.9339 - val_loss: 0.5132 - val_accuracy: 0.7944\n","Epoch 43/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1776 - accuracy: 0.9336 - val_loss: 0.4002 - val_accuracy: 0.8511\n","Epoch 44/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1766 - accuracy: 0.9343 - val_loss: 0.4516 - val_accuracy: 0.8253\n","Epoch 45/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1754 - accuracy: 0.9352 - val_loss: 0.4717 - val_accuracy: 0.8157\n","Epoch 46/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1751 - accuracy: 0.9353 - val_loss: 0.4193 - val_accuracy: 0.8401\n","Epoch 47/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1731 - accuracy: 0.9362 - val_loss: 0.4769 - val_accuracy: 0.8113\n","Epoch 48/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1719 - accuracy: 0.9359 - val_loss: 0.5106 - val_accuracy: 0.7951\n","Epoch 49/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1717 - accuracy: 0.9367 - val_loss: 0.4919 - val_accuracy: 0.8045\n","Epoch 50/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1704 - accuracy: 0.9361 - val_loss: 0.4554 - val_accuracy: 0.8224\n","Epoch 51/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1699 - accuracy: 0.9366 - val_loss: 0.3831 - val_accuracy: 0.8544\n","Epoch 52/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1684 - accuracy: 0.9372 - val_loss: 0.4182 - val_accuracy: 0.8381\n","Epoch 53/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1679 - accuracy: 0.9376 - val_loss: 0.3386 - val_accuracy: 0.8751\n","Epoch 54/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1662 - accuracy: 0.9383 - val_loss: 0.4168 - val_accuracy: 0.8381\n","Epoch 55/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1654 - accuracy: 0.9384 - val_loss: 0.3705 - val_accuracy: 0.8582\n","Epoch 56/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1653 - accuracy: 0.9390 - val_loss: 0.4279 - val_accuracy: 0.8331\n","Epoch 57/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1645 - accuracy: 0.9390 - val_loss: 0.4575 - val_accuracy: 0.8180\n","Epoch 58/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1639 - accuracy: 0.9391 - val_loss: 0.4258 - val_accuracy: 0.8330\n","Epoch 59/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1623 - accuracy: 0.9395 - val_loss: 0.3681 - val_accuracy: 0.8585\n","Epoch 60/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1622 - accuracy: 0.9393 - val_loss: 0.4110 - val_accuracy: 0.8407\n","Epoch 61/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1608 - accuracy: 0.9394 - val_loss: 0.3582 - val_accuracy: 0.8628\n","Epoch 62/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1609 - accuracy: 0.9396 - val_loss: 0.3956 - val_accuracy: 0.8444\n","Epoch 63/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1589 - accuracy: 0.9400 - val_loss: 0.4065 - val_accuracy: 0.8400\n","Epoch 64/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1592 - accuracy: 0.9410 - val_loss: 0.4204 - val_accuracy: 0.8342\n","Epoch 65/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1583 - accuracy: 0.9414 - val_loss: 0.4383 - val_accuracy: 0.8249\n","Epoch 66/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1576 - accuracy: 0.9415 - val_loss: 0.5283 - val_accuracy: 0.7852\n","Epoch 67/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1567 - accuracy: 0.9412 - val_loss: 0.4450 - val_accuracy: 0.8224\n","Epoch 68/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1572 - accuracy: 0.9408 - val_loss: 0.3700 - val_accuracy: 0.8547\n","Epoch 69/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1558 - accuracy: 0.9415 - val_loss: 0.3864 - val_accuracy: 0.8487\n","Epoch 70/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1544 - accuracy: 0.9418 - val_loss: 0.3751 - val_accuracy: 0.8527\n","Epoch 71/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1549 - accuracy: 0.9420 - val_loss: 0.3696 - val_accuracy: 0.8554\n","Epoch 72/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1535 - accuracy: 0.9428 - val_loss: 0.4225 - val_accuracy: 0.8312\n","Epoch 73/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1539 - accuracy: 0.9422 - val_loss: 0.2892 - val_accuracy: 0.8921\n","Epoch 74/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1526 - accuracy: 0.9429 - val_loss: 0.3858 - val_accuracy: 0.8465\n","Epoch 75/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1522 - accuracy: 0.9432 - val_loss: 0.3603 - val_accuracy: 0.8582\n","Epoch 76/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1517 - accuracy: 0.9428 - val_loss: 0.4344 - val_accuracy: 0.8255\n","Epoch 77/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1504 - accuracy: 0.9439 - val_loss: 0.5437 - val_accuracy: 0.7775\n","Epoch 78/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1507 - accuracy: 0.9430 - val_loss: 0.3965 - val_accuracy: 0.8413\n","Epoch 79/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1497 - accuracy: 0.9442 - val_loss: 0.4166 - val_accuracy: 0.8319\n","Epoch 80/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1505 - accuracy: 0.9437 - val_loss: 0.3688 - val_accuracy: 0.8540\n","Epoch 81/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1498 - accuracy: 0.9434 - val_loss: 0.4713 - val_accuracy: 0.8096\n","Epoch 82/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1484 - accuracy: 0.9444 - val_loss: 0.3360 - val_accuracy: 0.8690\n","Epoch 83/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1478 - accuracy: 0.9443 - val_loss: 0.2902 - val_accuracy: 0.8902\n","Epoch 84/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1475 - accuracy: 0.9449 - val_loss: 0.2889 - val_accuracy: 0.8903\n","Epoch 85/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1474 - accuracy: 0.9444 - val_loss: 0.4552 - val_accuracy: 0.8151\n","Epoch 86/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1461 - accuracy: 0.9456 - val_loss: 0.3757 - val_accuracy: 0.8509\n","Epoch 87/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1457 - accuracy: 0.9458 - val_loss: 0.3702 - val_accuracy: 0.8515\n","Epoch 88/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1465 - accuracy: 0.9453 - val_loss: 0.3773 - val_accuracy: 0.8496\n","Epoch 89/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1451 - accuracy: 0.9454 - val_loss: 0.4029 - val_accuracy: 0.8370\n","Epoch 90/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1442 - accuracy: 0.9462 - val_loss: 0.3521 - val_accuracy: 0.8610\n","Epoch 91/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1443 - accuracy: 0.9455 - val_loss: 0.3339 - val_accuracy: 0.8703\n","Epoch 92/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1436 - accuracy: 0.9458 - val_loss: 0.3936 - val_accuracy: 0.8416\n","Epoch 93/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1431 - accuracy: 0.9469 - val_loss: 0.3838 - val_accuracy: 0.8458\n","Epoch 94/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1423 - accuracy: 0.9462 - val_loss: 0.4109 - val_accuracy: 0.8334\n","Epoch 95/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1411 - accuracy: 0.9477 - val_loss: 0.2803 - val_accuracy: 0.8932\n","Epoch 96/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1410 - accuracy: 0.9460 - val_loss: 0.3761 - val_accuracy: 0.8497\n","Epoch 97/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1416 - accuracy: 0.9474 - val_loss: 0.2704 - val_accuracy: 0.8980\n","Epoch 98/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1405 - accuracy: 0.9468 - val_loss: 0.3712 - val_accuracy: 0.8512\n","Epoch 99/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1407 - accuracy: 0.9470 - val_loss: 0.2966 - val_accuracy: 0.8858\n","Epoch 100/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1398 - accuracy: 0.9472 - val_loss: 0.3384 - val_accuracy: 0.8670\n","Epoch 101/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1400 - accuracy: 0.9470 - val_loss: 0.3785 - val_accuracy: 0.8495\n","Epoch 102/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1393 - accuracy: 0.9477 - val_loss: 0.3207 - val_accuracy: 0.8752\n","Epoch 103/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1389 - accuracy: 0.9484 - val_loss: 0.3177 - val_accuracy: 0.8759\n","Epoch 104/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1388 - accuracy: 0.9477 - val_loss: 0.3889 - val_accuracy: 0.8426\n","Epoch 105/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1370 - accuracy: 0.9484 - val_loss: 0.3230 - val_accuracy: 0.8729\n","Epoch 106/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1373 - accuracy: 0.9484 - val_loss: 0.3334 - val_accuracy: 0.8685\n","Epoch 107/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1375 - accuracy: 0.9485 - val_loss: 0.4066 - val_accuracy: 0.8361\n","Epoch 108/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1374 - accuracy: 0.9488 - val_loss: 0.3893 - val_accuracy: 0.8405\n","Epoch 109/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1352 - accuracy: 0.9490 - val_loss: 0.2787 - val_accuracy: 0.8941\n","Epoch 110/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1354 - accuracy: 0.9494 - val_loss: 0.3068 - val_accuracy: 0.8808\n","Epoch 111/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1350 - accuracy: 0.9491 - val_loss: 0.2725 - val_accuracy: 0.8975\n","Epoch 112/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1348 - accuracy: 0.9492 - val_loss: 0.3748 - val_accuracy: 0.8493\n","Epoch 113/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1345 - accuracy: 0.9494 - val_loss: 0.3758 - val_accuracy: 0.8486\n","Epoch 114/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1337 - accuracy: 0.9496 - val_loss: 0.3059 - val_accuracy: 0.8811\n","Epoch 115/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1333 - accuracy: 0.9505 - val_loss: 0.3426 - val_accuracy: 0.8630\n","Epoch 116/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1331 - accuracy: 0.9496 - val_loss: 0.3752 - val_accuracy: 0.8490\n","Epoch 117/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1330 - accuracy: 0.9499 - val_loss: 0.3888 - val_accuracy: 0.8410\n","Epoch 118/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1316 - accuracy: 0.9506 - val_loss: 0.3455 - val_accuracy: 0.8625\n","Epoch 119/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1309 - accuracy: 0.9504 - val_loss: 0.3601 - val_accuracy: 0.8552\n","Epoch 120/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1311 - accuracy: 0.9508 - val_loss: 0.4260 - val_accuracy: 0.8254\n","Epoch 121/300\n","612/612 [==============================] - 12s 19ms/step - loss: 0.1297 - accuracy: 0.9515 - val_loss: 0.3630 - val_accuracy: 0.8541\n","Epoch 122/300\n","612/612 [==============================] - 11s 19ms/step - loss: 0.1302 - accuracy: 0.9511 - val_loss: 0.3704 - val_accuracy: 0.8510\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"qY8KLCc0n2tG","colab_type":"code","outputId":"07f51e4d-ee97-4796-8db6-7fb015a4ae3e","executionInfo":{"status":"ok","timestamp":1592061327150,"user_tz":-480,"elapsed":1777093,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":302}},"source":["# The change of training accuracy and validation accuracy\n","print(\"The training accuracy of section header model: {}%\".format(round(image_history.history[\"accuracy\"][-1]*100, 2)))\n","print(\"The validation accuracy of section header model: {}%\".format(round(image_history.history[\"val_accuracy\"][-1]*100, 2)))\n","\n","sns.set()\n","\n","image_acc = image_history.history[\"accuracy\"]\n","image_val_acc = image_history.history[\"val_accuracy\"]\n","\n","plt.plot(image_acc, \"b-\", label=\"acc\")\n","plt.plot(image_val_acc, \"r-\", label=\"val_acc\")\n","plt.legend()\n","plt.show()"],"execution_count":45,"outputs":[{"output_type":"stream","text":["The training accuracy of section header model: 95.11%\n","The validation accuracy of section header model: 85.1%\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]},{"cell_type":"markdown","metadata":{"id":"T8SC3FRR7dx5","colab_type":"text"},"source":["## Part5. Model Ensemble"]},{"cell_type":"code","metadata":{"id":"erKCYnB68sKd","colab_type":"code","colab":{}},"source":["def model_ensemble(models, model_input):\n","    outputs = [model.outputs[0] for model in models]\n","    y = keras.layers.Dense(8, activation=\"relu\")(outputs)\n","    y = keras.layers.Dense(1, activation=\"sigmoid\")(y)\n","    model = Model(model_input, y, name=\"model_ensemble\")\n","    return model"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"ERZbt8piJOP_","colab_type":"code","colab":{}},"source":["# Load models\n","header_dnn = header_model()\n","import_dnn = import_model()\n","image_cnn = image_model()"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"wraShhuAATBn","colab_type":"code","colab":{}},"source":["# combine the outputs of these three models\n","combinedInput = keras.layers.concatenate([header_dnn.output, import_dnn.output, image_cnn.output])\n","# Then add a Dense Layer in the end\n","x = keras.layers.Dense(16, activation=\"relu\")(combinedInput)\n","x = keras.layers.Dense(1, activation=\"sigmoid\")(x)\n","# The final ensemble model\n","model = keras.models.Model(inputs=[header_dnn.input, import_dnn.input, image_cnn.input], outputs=x)"],"execution_count":0,"outputs":[]},{"cell_type":"code","metadata":{"id":"pKusniGCLKHf","colab_type":"code","outputId":"a9feed7a-4cca-476b-c2a6-d5d19a42e9d8","executionInfo":{"status":"ok","timestamp":1592062146629,"user_tz":-480,"elapsed":2596504,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}},"colab":{"base_uri":"https://localhost:8080/","height":1000}},"source":["model.compile(optimizer=keras.optimizers.Adam(learning_rate=3e-4), \n","              loss=keras.losses.BinaryCrossentropy(), \n","              metrics=['accuracy'])\n","file_path = \"model_ensemble.hdf5\"\n","early_stopping = keras.callbacks.EarlyStopping(monitor='val_loss', patience=12)\n","checkpoint = keras.callbacks.ModelCheckpoint(file_path, monitor='loss', verbose=0, save_weights_only=True, save_best_only=True, mode='auto', period=1)\n","history = model.fit([header_X, import_X, image_X], y, epochs=300, callbacks=[early_stopping, checkpoint], validation_split=0.4, batch_size=80, shuffle=True)\n","model.save(file_path)"],"execution_count":49,"outputs":[{"output_type":"stream","text":["WARNING:tensorflow:`period` argument is deprecated. Please use `save_freq` to specify the frequency in number of batches seen.\n","Epoch 1/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.4307 - accuracy: 0.8357 - val_loss: 1.3783 - val_accuracy: 0.0000e+00\n","Epoch 2/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.2734 - accuracy: 0.8409 - val_loss: 0.8648 - val_accuracy: 0.7275\n","Epoch 3/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.1800 - accuracy: 0.9378 - val_loss: 0.5304 - val_accuracy: 0.8548\n","Epoch 4/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.1425 - accuracy: 0.9470 - val_loss: 0.4469 - val_accuracy: 0.8908\n","Epoch 5/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.1257 - accuracy: 0.9593 - val_loss: 0.4038 - val_accuracy: 0.9402\n","Epoch 6/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.1186 - accuracy: 0.9611 - val_loss: 0.4403 - val_accuracy: 0.9181\n","Epoch 7/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.1126 - accuracy: 0.9632 - val_loss: 0.3807 - val_accuracy: 0.9404\n","Epoch 8/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.1086 - accuracy: 0.9639 - val_loss: 0.3957 - val_accuracy: 0.9322\n","Epoch 9/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.1006 - accuracy: 0.9655 - val_loss: 0.3851 - val_accuracy: 0.9295\n","Epoch 10/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0938 - accuracy: 0.9679 - val_loss: 0.3599 - val_accuracy: 0.9306\n","Epoch 11/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0914 - accuracy: 0.9682 - val_loss: 0.3541 - val_accuracy: 0.9312\n","Epoch 12/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0888 - accuracy: 0.9696 - val_loss: 0.3305 - val_accuracy: 0.9325\n","Epoch 13/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0871 - accuracy: 0.9699 - val_loss: 0.3223 - val_accuracy: 0.9341\n","Epoch 14/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0857 - accuracy: 0.9708 - val_loss: 0.3848 - val_accuracy: 0.9318\n","Epoch 15/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0844 - accuracy: 0.9715 - val_loss: 0.2943 - val_accuracy: 0.9456\n","Epoch 16/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0839 - accuracy: 0.9718 - val_loss: 0.3134 - val_accuracy: 0.9454\n","Epoch 17/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0828 - accuracy: 0.9715 - val_loss: 0.3066 - val_accuracy: 0.9474\n","Epoch 18/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0816 - accuracy: 0.9723 - val_loss: 0.2952 - val_accuracy: 0.9401\n","Epoch 19/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0806 - accuracy: 0.9729 - val_loss: 0.3089 - val_accuracy: 0.9373\n","Epoch 20/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0792 - accuracy: 0.9732 - val_loss: 0.3373 - val_accuracy: 0.9329\n","Epoch 21/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0794 - accuracy: 0.9730 - val_loss: 0.3276 - val_accuracy: 0.9350\n","Epoch 22/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0778 - accuracy: 0.9737 - val_loss: 0.3055 - val_accuracy: 0.9368\n","Epoch 23/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0776 - accuracy: 0.9737 - val_loss: 0.2869 - val_accuracy: 0.9500\n","Epoch 24/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0770 - accuracy: 0.9739 - val_loss: 0.2941 - val_accuracy: 0.9456\n","Epoch 25/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0758 - accuracy: 0.9741 - val_loss: 0.2731 - val_accuracy: 0.9494\n","Epoch 26/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0764 - accuracy: 0.9736 - val_loss: 0.3034 - val_accuracy: 0.9460\n","Epoch 27/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0726 - accuracy: 0.9748 - val_loss: 0.2796 - val_accuracy: 0.9505\n","Epoch 28/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0692 - accuracy: 0.9762 - val_loss: 0.2567 - val_accuracy: 0.9535\n","Epoch 29/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0670 - accuracy: 0.9788 - val_loss: 0.2648 - val_accuracy: 0.9476\n","Epoch 30/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0635 - accuracy: 0.9821 - val_loss: 0.2743 - val_accuracy: 0.9480\n","Epoch 31/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0619 - accuracy: 0.9822 - val_loss: 0.2494 - val_accuracy: 0.9489\n","Epoch 32/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0605 - accuracy: 0.9827 - val_loss: 0.2690 - val_accuracy: 0.9479\n","Epoch 33/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0587 - accuracy: 0.9833 - val_loss: 0.2697 - val_accuracy: 0.9473\n","Epoch 34/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0578 - accuracy: 0.9834 - val_loss: 0.2179 - val_accuracy: 0.9544\n","Epoch 35/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0565 - accuracy: 0.9836 - val_loss: 0.2420 - val_accuracy: 0.9418\n","Epoch 36/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0570 - accuracy: 0.9832 - val_loss: 0.2391 - val_accuracy: 0.9536\n","Epoch 37/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0552 - accuracy: 0.9841 - val_loss: 0.2784 - val_accuracy: 0.9457\n","Epoch 38/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0549 - accuracy: 0.9838 - val_loss: 0.2372 - val_accuracy: 0.9526\n","Epoch 39/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0552 - accuracy: 0.9835 - val_loss: 0.2232 - val_accuracy: 0.9509\n","Epoch 40/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0535 - accuracy: 0.9841 - val_loss: 0.2622 - val_accuracy: 0.9471\n","Epoch 41/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0534 - accuracy: 0.9837 - val_loss: 0.2102 - val_accuracy: 0.9599\n","Epoch 42/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0533 - accuracy: 0.9840 - val_loss: 0.2287 - val_accuracy: 0.9548\n","Epoch 43/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0529 - accuracy: 0.9838 - val_loss: 0.2531 - val_accuracy: 0.9513\n","Epoch 44/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0521 - accuracy: 0.9845 - val_loss: 0.2597 - val_accuracy: 0.9466\n","Epoch 45/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0512 - accuracy: 0.9845 - val_loss: 0.2513 - val_accuracy: 0.9534\n","Epoch 46/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0512 - accuracy: 0.9846 - val_loss: 0.2342 - val_accuracy: 0.9550\n","Epoch 47/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0514 - accuracy: 0.9846 - val_loss: 0.2409 - val_accuracy: 0.9513\n","Epoch 48/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0511 - accuracy: 0.9844 - val_loss: 0.2432 - val_accuracy: 0.9519\n","Epoch 49/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0506 - accuracy: 0.9848 - val_loss: 0.2406 - val_accuracy: 0.9497\n","Epoch 50/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0501 - accuracy: 0.9847 - val_loss: 0.2529 - val_accuracy: 0.9512\n","Epoch 51/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0499 - accuracy: 0.9853 - val_loss: 0.2289 - val_accuracy: 0.9559\n","Epoch 52/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0497 - accuracy: 0.9852 - val_loss: 0.2242 - val_accuracy: 0.9565\n","Epoch 53/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0499 - accuracy: 0.9854 - val_loss: 0.2096 - val_accuracy: 0.9633\n","Epoch 54/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0493 - accuracy: 0.9852 - val_loss: 0.2426 - val_accuracy: 0.9507\n","Epoch 55/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0497 - accuracy: 0.9852 - val_loss: 0.2291 - val_accuracy: 0.9559\n","Epoch 56/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0487 - accuracy: 0.9855 - val_loss: 0.2583 - val_accuracy: 0.9515\n","Epoch 57/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0486 - accuracy: 0.9854 - val_loss: 0.2533 - val_accuracy: 0.9535\n","Epoch 58/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0480 - accuracy: 0.9859 - val_loss: 0.2222 - val_accuracy: 0.9603\n","Epoch 59/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0494 - accuracy: 0.9854 - val_loss: 0.2446 - val_accuracy: 0.9551\n","Epoch 60/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0493 - accuracy: 0.9851 - val_loss: 0.2461 - val_accuracy: 0.9553\n","Epoch 61/300\n","612/612 [==============================] - 13s 20ms/step - loss: 0.0477 - accuracy: 0.9856 - val_loss: 0.2213 - val_accuracy: 0.9591\n","Epoch 62/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0481 - accuracy: 0.9856 - val_loss: 0.2404 - val_accuracy: 0.9508\n","Epoch 63/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0477 - accuracy: 0.9858 - val_loss: 0.2347 - val_accuracy: 0.9578\n","Epoch 64/300\n","612/612 [==============================] - 13s 21ms/step - loss: 0.0476 - accuracy: 0.9855 - val_loss: 0.2184 - val_accuracy: 0.9616\n","Epoch 65/300\n","612/612 [==============================] - 12s 20ms/step - loss: 0.0479 - accuracy: 0.9858 - val_loss: 0.2185 - val_accuracy: 0.9599\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"0BY3qCtyefRG","colab_type":"code","colab":{"base_uri":"https://localhost:8080/","height":302},"outputId":"0391237f-dda5-4890-a5ce-1223606de4af","executionInfo":{"status":"ok","timestamp":1592062146631,"user_tz":-480,"elapsed":2596493,"user":{"displayName":"郭宇雋","photoUrl":"","userId":"01461113603099005024"}}},"source":["# The change of training accuracy and validation accuracy\n","print(\"The training accuracy of section header model: {}%\".format(round(history.history[\"accuracy\"][-1]*100, 2)))\n","print(\"The validation accuracy of section header model: {}%\".format(round(history.history[\"val_accuracy\"][-1]*100, 2)))\n","\n","sns.set()\n","\n","acc = history.history[\"accuracy\"]\n","val_acc = history.history[\"val_accuracy\"]\n","\n","plt.plot(acc, \"b-\", label=\"acc\")\n","plt.plot(val_acc, \"r-\", label=\"val_acc\")\n","plt.legend()\n","plt.show()"],"execution_count":50,"outputs":[{"output_type":"stream","text":["The training accuracy of section header model: 98.58%\n","The validation accuracy of section header model: 95.99%\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[]}}]}]}