ML-Corrupted.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ` > `[`@qwqoro`](https://github.com/qwqoro)`: Welcome! `\n",
    "<br>\n",
    "\n",
    "        // I created this out of interest in the summer of 2019 at the age of 15\n",
    "        // Resources, links and libraries were updated in the spring of 2022"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-info\">\n",
    "<b>Machine Learning level:</b> Script Kiddie\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Contents:\n",
    "\n",
    "- <h3><a href=\"#💡-Idea\">💡: Idea</a></h3>\n",
    "    \n",
    "- <h3><a href=\"#📋-Data\">📋: Data</a></h3>\n",
    "- <h3><a href=\"#✨-Preparations\">✨: Preparations</a></h3>\n",
    "\n",
    "    Anatomy of each file format; Making files corrupted\n",
    "    - [Images](#Images): `BMP`,`JPEG`,`PNG`\n",
    "    - [Videos](#Videos): `AVI`\n",
    "    - [Audios](#Audios): `WAV`,`MP3`\n",
    "    - [Executables](#Executables): `PE`\n",
    "    \n",
    "- <h3><a href=\"#🔎-Features\">🔎: Features</a></h3>\n",
    "\n",
    "    Process of feature engineering\n",
    "    - [Thoughts](#Thoughts)\n",
    "    - [Construction](#Construction)\n",
    "    \n",
    "- <h3><a href=\"#🤖-Training\">🤖: Training</a></h3>\n",
    "\n",
    "    - [Random Forest](#Random-Forest)\n",
    "    - [K-Nearest Neighbors](#K-Nearest-Neighbors)\n",
    "    - [CatBoost](#CatBoost)\n",
    "    - [XGBoost](#XGBoost)\n",
    "    - [LightGBM](#LightGBM)\n",
    "- <h3><a href=\"#📊-Comparison\">📊: Comparison</a></h3>\n",
    "- <h3><a href=\"#✍-Testing\">✍: Testing</a></h3>\n",
    "- <h3><a href=\"#📚-Resources\">📚: Resources</a></h3>\n",
    "<br>\n",
    "<br>\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 💡 Idea\n",
    "_[> Back to contents](#Contents:)_\n",
    "<br>\n",
    "\n",
    "Solving CTF tasks led me to corrupted files at some point. When they lack distinctive features, such as signatures, most of the tools identify them as \"`Text files`\" or \"`Data`\", which is not useful at all, to be honest.\n",
    "\n",
    "To proceed, it would be great to know their types, at least. Even though some files might be too broken to be fixed, unless you know the exact file format, which parts are missing and which part you have got before your eyes, I believe that guessing types may still be useful in some cases, e.g. criminal investigations — it might be important to find remains of what used to be evidences and analyze them thoroughly. Thus, my aim is to build models that would guess types of broken files.\n",
    "\n",
    "<br>\n",
    "\n",
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 📋 Data\n",
    "_[> Back to contents](#Contents:)_\n",
    "<br>\n",
    "\n",
    "<div class=\"alert alert-block alert-warning\">\n",
    "⚠️ <b>Every piece of data belongs to its author(s)!</b> ⚠️<br>\n",
    "⚠️ <b>All copyrights belongs to the original owners!</b> ⚠️<br>\n",
    "The sources are listed here and credited in the <a href=\"#📚-Resources\">\"Resources\"</a> section.\n",
    "</div><br>\n",
    "\n",
    "    // I use data for non-commercial purposes, as a hobbyist.\n",
    "    // I do not distribute data.\n",
    "<br>\n",
    "\n",
    "- 🎨 **Images** (Raster)\n",
    "    - (Uncompressed) `BMP`: [General-100 Dataset](http://mmlab.ie.cuhk.edu.hk/projects/FSRCNN.html) `[1]` + random files gathered from the Internet\n",
    "    - (Lossy compression) `JPEG`: [UTKFace: Large Scale Face Dataset](https://susanqq.github.io/UTKFace/) `[2]`\n",
    "    - (Lossless compression) `PNG`: random files gathered from the Internet\n",
    "- 🎬 **Videos**\n",
    "    - `AVI`: [HMDB: A Large Video Database for Human Motion Recognition](http://serre-lab.clps.brown.edu/resource/hmdb-a-large-human-motion-database/) `[3]`\n",
    "- 🎧 **Audios**\n",
    "    - (Uncompressed) `WAV`: [FSDnoisy18k](http://www.eduardofonseca.net/FSDnoisy18k/) `[4]`\n",
    "    - (Lossy compression) `MP3`: [Mozilla Common Voice Dataset](https://commonvoice.mozilla.org/en/datasets)\n",
    "- ⚙️ **Executables**\n",
    "    - Portable Executables: [Microsoft Malware Classification Challenge](https://www.kaggle.com/c/malware-classification/) `[5]`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000 files of BMP format \t(Images)\n",
      "1000 files of JPG format \t(Images)\n",
      "1000 files of PNG format \t(Images)\n",
      "3000 files of AVI format \t(Videos)\n",
      "1500 files of WAV format \t(Audios)\n",
      "1500 files of MP3 format \t(Audios)\n",
      "3000 files of PE format \t(Executables)\n",
      "--------------------------------\n",
      "Total number of files:  12000\n"
     ]
    }
   ],
   "source": [
    "# Count files of each file format\n",
    "\n",
    "from os import listdir\n",
    "from os.path import isfile\n",
    "\n",
    "folders = [ \"Images_BMP\", \"Images_JPG\", \"Images_PNG\",\"Videos_AVI\",\"Audios_WAV\",\"Audios_MP3\",\"Executables_PE\" ]\n",
    "total = 0\n",
    "\n",
    "for Type in folders:\n",
    "    path = f\"data/{Type}/\"\n",
    "    Type = Type.split('_')\n",
    "    \n",
    "    files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "    num = len(files)\n",
    "    total += num\n",
    "    \n",
    "    print(f\"{num} files of {Type[1]} format \\t({Type[0]})\")\n",
    "    \n",
    "\n",
    "print('-' * 32)\n",
    "print(\"Total number of files: \", total)"
   ]
  },
  {
   "attachments": {
    "image.png": {
     "image/png": "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"
    }
   },
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![](attachment:image.png)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ✨ Preparations\n",
    "_[> Back to contents](#Contents:)_\n",
    "\n",
    "<br>\n",
    "\n",
    "In this part I break down each file format.\n",
    "<br>By that I mean **describe their structure** and then **make them corrupted** to extract features & train models.\n",
    "<br>\n",
    "\n",
    "<div class=\"alert alert-block alert-warning\">\n",
    "⚠️ <b>Every image belongs to its author(s)!</b> ⚠️<br>\n",
    "⚠️ <b>All copyrights belongs to the original owners!</b> ⚠️<br>\n",
    "The sources are listed here and credited in the <a href=\"#📚-Resources\">\"Resources\"</a> section.\n",
    "</div><br>\n",
    "\n",
    "    // I use images for demonstration purposes; I did not make any changes to these images."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm import tqdm                    # to display progress\n",
    "from binascii import hexlify             # to view hexadecimal representation of binary data\n",
    "from binascii import unhexlify\n",
    "from re import finditer                  # to find all occurrences of a byte sequence in file"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Images\n",
    "- <h3>BMP</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "b'424df208130000000000360000002800000086020000830200000100180000000000bc081300c40e0000c40e000000000000000000001d0c0f1d0c0f1d0c0f1d0c0f1d0c0f1d0c0f1d0c0f1d0c0f1e0d101e0d101e0d101f0e111f0e11200f12200f12200f121f0e111f0e111f0e11200f12200f12211013211013211013221114221114221114221114221114221114221114221114221114221114221114221114221114221114221114221114200f12200f12200f121f0e111f0e111e0d101e0d101e0d101d0c0f1d0c0f1d0c0f1c0b0e1c0b0e1b0a0d1b0a0d1b0a0d1b0a0d1b0a0d1b0a0d1b0a0d1b0a0d1b0a0d1b0a0d1b0a0d1d080b1d080b1d080b1c'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hexlify(open(\"data/Images_BMP/0.bmp\", 'rb').read())[:512]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Analysis:\n",
    "\n",
    "|HEX|DEC|Meaning|BE/LE|Unit|\n",
    "|-:|:-|:-|:-:|:-:|\n",
    "|424D|66, 77 (\"`BM`\")|signature|(big-endian)||\n",
    "|F2081300|1247474‬|size of file|(little-endian)|bytes|\n",
    "|00000000|0|reserved|||\n",
    "|36000000|**54**|offset to the start of the bitmap data|(little-endian)|bytes|\n",
    "|28000000|40|size of information header|(little-endian)|bytes|\n",
    "|86020000|646|width|(little-endian)|pixels|\n",
    "|83020000|643|height|(little-endian)|pixels|\n",
    "|0100|1|number of planes|(little-endian)||\n",
    "|1800|24 (24bit RGB)|number of bits/pixel|(little-endian)||\n",
    "|00000000|0 (None)|compression type|||\n",
    "|BC081300|1247420|size of image data|(little-endian)|bytes|\n",
    "|C40E0000|3780‬|horizontal resolution|(little-endian)|pixels/meter|\n",
    "|C40E0000|3780|vertical resolution|(little-endian)|pixels/meter|\n",
    "|00000000|0|number of used colors OR zero|||\n",
    "|00000000|0|number of important colors OR zero|||"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "file: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 1000/1000 [01:01<00:00, 16.31it/s]\n"
     ]
    }
   ],
   "source": [
    "# Remove the first 54 bytes\n",
    "\n",
    "path = \"data/Images_BMP/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "\n",
    "for file in tqdm(files, desc='file'):\n",
    "    hexfile = open(path + file, \"rb\").read()\n",
    "    newfile = hexfile[54:]\n",
    "    \n",
    "    open(f\"data_edited/{file}\", 'wb').write(newfile)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- <h3>JPEG</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "b'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'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hexlify(open(\"data/Images_JPG/0.jpg\", 'rb').read())[:512]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "JPEG markers:\n",
    "\n",
    "|HEX|Meaning|\n",
    "|-:|:-|\n",
    "|FFD8|start of image|\n",
    "|FFC0|start of frame (baseline DCT-based JPEG)|\n",
    "|FFC1|start of frame (extended DCT-based JPEG)|\n",
    "|FFC2|start of frame (progressive DCT-based JPEG)|\n",
    "|FFC3|start of frame (lossless sequential)|\n",
    "|FFC4|define Huffman table(s)|\n",
    "|FFC5|start of frame (differential sequential DCT-based JPEG)|\n",
    "|FFC6|start of frame (differential progressive DCT-based JPEG)|\n",
    "|FFC7|start of frame (differential lossless sequential)|\n",
    "|FFC9|start of frame (extended sequential DCT, arithmetic coding)|\n",
    "|FFCA|start of frame (progressive DCT-based JPEG, arithmetic coding)|\n",
    "|FFCB|start of frame (lossless sequential, arithmetic coding)|\n",
    "|FFCC|define arithmetic coding|\n",
    "|FFCD|start of frame (differential sequential DCT-based JPEG, arithmetic coding)|\n",
    "|FFCE|start of frame (differential progressive DCT-based JPEG, arithmetic coding)|\n",
    "|FFCF|start of frame (differential lossless sequential, arithmetic coding)|\n",
    "|FFDB|define Quantization table(s)|\n",
    "|FFDD|define restart interval|\n",
    "|FFDA|start of scan|\n",
    "|FFDn|restart (`n` = 0..7)|\n",
    "|FFEn|application-specific (`n` = 0..7)|\n",
    "|FFFE|comment|\n",
    "|FFD9|end of image|"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "file: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 1000/1000 [00:20<00:00, 49.70it/s]\n"
     ]
    }
   ],
   "source": [
    "# Leave only bytes between SOS (Start Of Scan) and EOI (End Of Image) markers\n",
    "\n",
    "path = \"data/Images_JPG/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "\n",
    "for file in tqdm(files, desc=\"file\"):\n",
    "    hexfile = open(path + file, 'rb').read()\n",
    "    \n",
    "    SOS = hexfile.find(b\"\\xFF\\xDA\")\n",
    "    EOI = hexfile.find(b\"\\xFF\\xD9\")\n",
    "    \n",
    "    newfile = hexfile[SOS + 4 : EOI]\n",
    "    open(f\"data_edited/{file}\", 'wb').write(newfile)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- <h3>PNG</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "b'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'"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hexlify(open(\"data/Images_PNG/0.png\", 'rb').read())[:512]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "PNG chunks:\n",
    "\n",
    "||Meaning|\n",
    "|-:|:-|\n",
    "|IHDR|signature|\n",
    "|PLTE|palette|\n",
    "|IDAT|image itself|\n",
    "|IEND|end of image|\n",
    "|bKGD|default background color|\n",
    "|cHRM|chromaticity coordinates|\n",
    "|dSIG|digital signatures|\n",
    "|eXIf|exif metadata|\n",
    "|gAMA|gamma|\n",
    "|hIST|histogram|\n",
    "|iCCP|ICC color profile|\n",
    "|iTXt|keyword and UTF-8 text|\n",
    "|pHYs|intended pixel size (or pixel aspect ratio)|\n",
    "|sBIT|significant bits, color-accuracy|\n",
    "|sPLT|palette if the full range of colors is unavailable|\n",
    "|sRGB|indicates that sRGB color space is used|\n",
    "|sTER|stereo-image indicator|\n",
    "|tEXt|text|\n",
    "|tIME|time of last modification|\n",
    "|tRNS|transparency|\n",
    "|zTXt|compressed text|"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "file: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 1000/1000 [00:46<00:00, 21.37it/s]\n"
     ]
    }
   ],
   "source": [
    "# Remove everything except the `IDAT` chunk contents\n",
    "\n",
    "path = \"data/Images_PNG/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "\n",
    "for file in tqdm(files, desc=\"file\"):\n",
    "    hexfile = open(path + file, 'rb').read()\n",
    "    newfile = []\n",
    "    idats = finditer(b\"\\x49\\x44\\x41\\x54\", hexfile)\n",
    "    \n",
    "    for marker in idats:\n",
    "        section_length = int(hexlify(hexfile[marker.start()-4 : marker.start()]), 16)\n",
    "        indexes = (marker.end(), marker.end() + section_length)\n",
    "        newfile.append(hexfile[indexes[0]:indexes[1]])\n",
    "        \n",
    "    open(f\"data_edited/{file}\", 'wb').write(b''.join(newfile))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-success\" style=\"margin: auto; width: 256px; padding: 16px 32px\">\n",
    "<b>Done: </b> Images ✅\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Videos\n",
    "- <h3>AVI</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "b'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'"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hexlify(open(\"data/Videos_AVI/0.avi\", \"rb\").read())[:512]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "AVI chunks:\n",
    "\n",
    "||Meaning|\n",
    "|-:|:-|\n",
    "|LIST hdrl|header|\n",
    "|LIST movi|actual data|\n",
    "|LIST idx1|indexes|\n",
    "|strh|stream header|\n",
    "|strd|stream-header data|\n",
    "|strf|stream format|\n",
    "|strn|stream name|"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "file: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 3000/3000 [01:06<00:00, 45.01it/s]\n"
     ]
    }
   ],
   "source": [
    "# Remove everything except the `LIST movi` chunk contents\n",
    "\n",
    "path = \"data/Videos_AVI/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "\n",
    "for file in tqdm(files, desc=\"file\"):\n",
    "    hexfile = open(path + file, 'rb').read()\n",
    "    newfile = []\n",
    "    lists = finditer(b\"\\x6D\\x6F\\x76\\x69\", hexfile)\n",
    "    \n",
    "    for marker in lists:\n",
    "        indexes = (marker.end()+2, hexfile[marker.end():].find(b\"\\x4C\\x49\\x53\\x54\"))\n",
    "        newfile.append(hexfile[indexes[0]:indexes[1]])\n",
    "        \n",
    "    open(f\"data_edited/{file}\", 'wb').write(b''.join(newfile))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-success\" style=\"margin: auto; width: 256px; padding: 16px 32px\">\n",
    "<b>Done: </b> Videos ✅\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Audios\n",
    "- <h3>WAV</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "b'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'"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hexlify(open(\"data/Audios_WAV/0.wav\", \"rb\").read())[:512]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`[6]` [WAV structure](https://www.researchgate.net/figure/Format-of-WAV-document_tbl3_281444376):\n",
    "<img src=\"https://www.researchgate.net/profile/Haviluddin-H/publication/281444376/figure/tbl3/AS:667902516658179@1536251777729/Format-of-WAV-document.png\" style=\"width: 600px; height: auto\" alt=\"WAV structure\"/>\n",
    "\n",
    "WAV chunks:\n",
    "\n",
    "||Meaning|\n",
    "|-:|:-|\n",
    "|fmt|format of the data|\n",
    "|fact|number of samples|\n",
    "|data|data|"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "file: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 1500/1500 [01:36<00:00, 15.55it/s]\n"
     ]
    }
   ],
   "source": [
    "# Remove everything except the `data` chunk contents\n",
    "\n",
    "path = \"data/Audios_WAV/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "\n",
    "for file in tqdm(files, desc=\"file\"):\n",
    "    hexfile = open(path + file, 'rb').read()\n",
    "    newfile = []\n",
    "    datas = finditer(b\"\\x64\\x61\\x74\\x61\", hexfile)\n",
    "    \n",
    "    for marker in datas:\n",
    "        section_length = int(hexlify(hexfile[marker.end() : marker.end()+4]), 16)\n",
    "        indexes = (marker.end(), marker.end() + section_length)\n",
    "        newfile.append(hexfile[indexes[0]:indexes[1]])\n",
    "        \n",
    "    open(f\"data_edited/{file}\", 'wb').write(b''.join(newfile))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- <h3>MP3</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "b'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'"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hexlify(open(\"data/Audios_MP3/0.mp3\", \"rb\").read())[:512]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`[7]`   [MP3 structure](https://commons.wikimedia.org/wiki/File:Mp3filestructure.svg):\n",
    "- License: [Attribution 2.5 Generic](https://creativecommons.org/licenses/by/2.5/deed.en)\n",
    "- Authors: [Original version by Kim Meyrick](https://en.wikipedia.org/wiki/User:Kim_Meyrick) + [Later version by Stannered](https://commons.wikimedia.org/wiki/User:Stannered) + [Current version by Mik](https://commons.wikimedia.org/wiki/User:Mik)\n",
    "\n",
    "    <img src=\"https://upload.wikimedia.org/wikipedia/commons/0/01/Mp3filestructure.svg\" style=\"width: 900px; height: auto\" alt=\"MP3 structure\"/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "file: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 1500/1500 [00:24<00:00, 61.83it/s]\n"
     ]
    }
   ],
   "source": [
    "# Remove everything except the MP3 frames (start with `FFFx`) contents\n",
    "\n",
    "path = \"data/Audios_MP3/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "\n",
    "for file in tqdm(files, desc=\"file\"):\n",
    "    hexfile = open(path + file, 'rb').read()\n",
    "    newfile = []\n",
    "    frames = list(finditer(b\"\\xFF[\\xF0|\\xF1|\\xF2|\\xF3|\\xF4|\\xF5|\\xF6|\\xF7|\\xFA|\\xFB|\\xFC|\\xFD|\\xFE|\\xFF]\", hexfile))\n",
    "\n",
    "    for marker in range(len(frames)-1):\n",
    "        indexes = (frames[marker].end(), frames[marker+1].start())\n",
    "        newfile.append(hexfile[indexes[0]:indexes[1]])\n",
    "\n",
    "    open(f\"data_edited/{file}\", 'wb').write(b''.join(newfile))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-success\" style=\"margin: auto; width: 256px; padding: 16px 32px\">\n",
    "<b>Done: </b> Audios ✅\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Executables\n",
    "- <h3>PE</h3>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "b'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'"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "open(\"data/Executables_PE/0.bytes\", 'rb').read().lower()[:512]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`[8]`   [PE (x32) structure](https://commons.wikimedia.org/wiki/File:Portable_Executable_32_bit_Structure_in_SVG_fixed.svg):\n",
    "- License: [Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/deed.en)\n",
    "- Authors: Original version by ByteBiter + [Later version by Benhut1](https://commons.wikimedia.org/wiki/User:Benhut1) + [Current version by Finnusertop](https://commons.wikimedia.org/wiki/User:Finnusertop)\n",
    "\n",
    "    <img src=\"https://upload.wikimedia.org/wikipedia/commons/1/1b/Portable_Executable_32_bit_Structure_in_SVG_fixed.svg\" style=\"height: 800px; width: auto\" alt=\"PE (x32) structure\"/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "file: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 3000/3000 [03:04<00:00, 16.27it/s]\n"
     ]
    }
   ],
   "source": [
    "# All of the files are already sterile\n",
    "\n",
    "path = \"data/Executables_PE/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "\n",
    "for file in tqdm(files, desc=\"file\"):\n",
    "    hexfile = open(path + file, 'rb').read()\n",
    "    hexfile = unhexlify(hexfile)\n",
    "    \n",
    "    file = file.replace(\"bytes\",\"exe\")\n",
    "    open(f\"data_edited/{file}\", 'wb').write(hexfile)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-success\" style=\"margin: auto; width: 256px; padding: 16px 32px\">\n",
    "<b>Done: </b> PE & everything ✅\n",
    "</div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 🔎 Features\n",
    "_[> Back to contents](#Contents:)_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Thoughts\n",
    "0. file size in bytes\n",
    "1. \\* bytes in file: `all / commonest / rarest`\n",
    "2. \\* bytes in each part of file (file is split into K parts): `all / commonest / rarest`\n",
    "3. \\* N-grams (stride = 1): `commonest / rarest`\n",
    "4. \\* N-grams (stride = N): `commonest / rarest`\n",
    "5. \\* N-grams (stride = 1) in each part of file (file is split into K parts): `commonest / rarest`\n",
    "6. \\* N-grams (stride = N) in each part of file (file is split into K parts): `commonest / rarest`\n",
    "\n",
    "<br>\n",
    "Resulting DataFrames:\n",
    "\n",
    "|idea|number of features|\n",
    "|:-|:-:|\n",
    "|10 most common bytes in each part of file split into 3 parts|30|\n",
    "|20 most common byte 4-grams (stride = 1 byte)|20|\n",
    "|20 most common byte 6-grams (stride = 2 bytes)|20|\n",
    "|20 most common byte 8-grams (stride = 4 bytes)|20|\n",
    "|10 most common byte 4-grams (stride = 1 byte) in each part of file split into 3 parts|30|\n",
    "|10 most common byte 6-grams (stride = 2 bytes) in each part of file split into 3 parts|30|\n",
    "|10 most common byte 8-grams (stride = 4 bytes) in each part of file split into 3 parts|30|\n",
    "\n",
    "**However**:<br>\n",
    "- There are cases when the #1 most popular object is encountered 1000 times in an interval, #2 - 900 times, #3 - only 100 times. That is why it makes sense to rearrange every array of the most popular objects so that it would also depict popularity of each one of its objects among the other objects within it. That is done further with use of a custom `fill` function.<br>\n",
    "<ins>Example</ins>:<br>\n",
    "```\n",
    "top = [(\"1\",1000), (\"2\",900), (\"3\",100), (\"4\",10), (\"5\",9)]\n",
    "// \"1\" → ~50%, \"2\" → ~45%, \"3\" → ~5%, \"4\" → ~0.5%, \"5\" → ~0.45%\n",
    "fill(top, 5) → \"1\", \"1\", \"1\", \"2\", \"2\"\n",
    "fill(top, 10) → \"1\", \"1\", \"1\", \"1\", \"1\", \"2\", \"2\", \"2\", \"2\", \"2\"\n",
    "fill(top, 20) → \"1\", \"1\", \"1\", \"1\", \"1\", \"1\", \"1\", \"1\", \"1\", \"1\", \"2\", \"2\", \"2\", \"2\", \"2\", \"2\", \"2\", \"2\", \"2\", \"3\"\n",
    "```\n",
    "<br>\n",
    "\n",
    "- In a function called `calc` that is introduced later, hexadecimal values of bytes are converted into decimal numbers. That is done because bytes and byte sequences that belong to the same category _(digits, latin alphabet, cyrillic alphabet, special characters, emoji, ...)_ are usually close in numeric value.<br>\n",
    "<ins>Example</ins>:<br>\n",
    "```\n",
    "CEB1 → 52913 → (UTF-8) α; CEB2 → 52914 → (UTF-8) β\n",
    "D0AB → 53419 → (UTF-8) Ы; D0AC → 53420 → (UTF-8) Ь\n",
    "F09FA6BE → 4036994750 → (UTF-8) 🦾; F09FA6BF → 4036994751 → (UTF-8) 🦿\n",
    "F09FAB80 → 4036995968 → (UTF-8) 🫀; F09FAB81 → 4036995969 → (UTF-8) 🫁\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Construction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from math import ceil\n",
    "from collections import Counter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define names of columns\n",
    "\n",
    "labels = [\"Type\",\"Format\"]\n",
    "\n",
    "NGrams = [f\"top{x}\" for x in range(1,21)]\n",
    "\n",
    "Split  = [f\"top1.{x}\" for x in range(1,11)]\n",
    "Split += [f\"top2.{x}\" for x in range(1,11)]\n",
    "Split += [f\"top3.{x}\" for x in range(1,11)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create DataFrames for each case\n",
    "#   // GN → N-Grams\n",
    "#   //  S → file is split into parts\n",
    "\n",
    "S   = pd.DataFrame(columns = labels + Split)\n",
    "G4  = pd.DataFrame(columns = labels + NGrams)\n",
    "G6  = pd.DataFrame(columns = labels + NGrams)\n",
    "G8  = pd.DataFrame(columns = labels + NGrams)\n",
    "G4S = pd.DataFrame(columns = labels + Split)\n",
    "G6S = pd.DataFrame(columns = labels + Split)\n",
    "G8S = pd.DataFrame(columns = labels + Split)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Identify type depending on extension\n",
    "\n",
    "def get_type(frmt):\n",
    "    if frmt in [\"bmp\",\"jpg\",\"png\"]: return \"Image\"\n",
    "    elif     frmt in [\"wav\",\"mp3\"]: return \"Audio\"\n",
    "    elif             frmt == \"avi\": return \"Video\"\n",
    "    elif             frmt == \"exe\": return \"Executable\"\n",
    "    else:                           return  None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_ngrams(Bytes, params):\n",
    "    N, stride = params[0], params[1]\n",
    "    \n",
    "    return [ ''.join(Bytes[i:i+N]) for i in range(0, len(Bytes)-N+1, stride) ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fill(top, aimSize):\n",
    "    topNew = []\n",
    "    totalSize = sum([el[1] for el in top])\n",
    "    \n",
    "    for el in top:\n",
    "        importance = ceil(aimSize * el[1]/totalSize)\n",
    "        topNew += [(el[0], 0)] * importance\n",
    "        \n",
    "    return topNew[:aimSize]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate everything according to the idea (`mode`)\n",
    "\n",
    "def calc(Bytes, part1, part2, part3, frmt, mode):\n",
    "    Result = []\n",
    "    \n",
    "    if mode == \"S\":\n",
    "        top1 = Counter(part1).most_common()[:10]\n",
    "        top2 = Counter(part2).most_common()[:10]\n",
    "        top3 = Counter(part3).most_common()[:10]\n",
    "        top1 = sorted(top1, key=lambda el: (el[1], el[0]))[::-1]\n",
    "        top2 = sorted(top2, key=lambda el: (el[1], el[0]))[::-1]\n",
    "        top3 = sorted(top3, key=lambda el: (el[1], el[0]))[::-1]\n",
    "        \n",
    "        top1 = fill(top1, 10)\n",
    "        top2 = fill(top2, 10)\n",
    "        top3 = fill(top3, 10)\n",
    "        \n",
    "        top1 = [ int(Set[0], 16) for Set in top1 ]\n",
    "        top2 = [ int(Set[0], 16) for Set in top2 ]\n",
    "        top3 = [ int(Set[0], 16) for Set in top3 ]\n",
    "        Result = top1 + top2 + top3\n",
    "        \n",
    "        \n",
    "    elif mode in [\"G4\",\"G6\",\"G8\"]:\n",
    "        params = {\"G4\": (4, 1),\n",
    "                  \"G6\": (6, 2),\n",
    "                  \"G8\": (8, 4)}\n",
    "        \n",
    "        ngrams = get_ngrams(Bytes, params[mode])\n",
    "        top = Counter(ngrams).most_common()[:20]\n",
    "        top = sorted(top, key=lambda el: (el[1], el[0]))[::-1]\n",
    "\n",
    "        top = fill(top, 20)\n",
    "            \n",
    "        top = [ int(Set[0], 16) for Set in top ]\n",
    "        Result = top\n",
    "        \n",
    "        \n",
    "    elif mode in [\"G4S\",\"G6S\",\"G8S\"]:\n",
    "        params = {\"G4S\": (4, 1),\n",
    "                  \"G6S\": (6, 2),\n",
    "                  \"G8S\": (8, 4)}\n",
    "        \n",
    "        ngrams1 = get_ngrams(part1, params[mode])\n",
    "        ngrams2 = get_ngrams(part2, params[mode])\n",
    "        ngrams3 = get_ngrams(part3, params[mode])\n",
    "        top1 = Counter(ngrams1).most_common()[:10]\n",
    "        top2 = Counter(ngrams2).most_common()[:10]\n",
    "        top3 = Counter(ngrams3).most_common()[:10]\n",
    "        top1 = sorted(top1, key=lambda el: (el[1], el[0]))[::-1]\n",
    "        top2 = sorted(top2, key=lambda el: (el[1], el[0]))[::-1]\n",
    "        top3 = sorted(top3, key=lambda el: (el[1], el[0]))[::-1]\n",
    "\n",
    "        top1 = fill(top1, 10)\n",
    "        top2 = fill(top2, 10)\n",
    "        top3 = fill(top3, 10)\n",
    "            \n",
    "        top1 = [ int(Set[0], 16) for Set in top1 ]\n",
    "        top2 = [ int(Set[0], 16) for Set in top2 ]\n",
    "        top3 = [ int(Set[0], 16) for Set in top3 ]\n",
    "        Result = top1 + top2 + top3\n",
    "    \n",
    "\n",
    "    return Result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "File: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 12000/12000 [5:40:06<00:00,  1.70s/it]\n"
     ]
    }
   ],
   "source": [
    "path  = \"data_edited/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "        \n",
    "for file in tqdm(files, desc='File'):\n",
    "    \n",
    "    frmt    = file.split('.')[1]                                          # format\n",
    "    hexfile = hexlify(open(path + file, 'rb').read()).decode()            # hex representation\n",
    "    Bytes = [ hexfile[i:i+2] for i in range(0, len(hexfile)-1, 2) ]       # \n",
    "    \n",
    "    partSize = len(Bytes) // 3\n",
    "    part1 = Bytes[            :   partSize]\n",
    "    part2 = Bytes[   partSize : 2*partSize]\n",
    "    part3 = Bytes[ 2*partSize :           ]\n",
    "\n",
    "    S   =   S.append( pd.DataFrame([[ get_type(frmt), frmt ] + calc(Bytes, part1, part2, part3, frmt, mode=\"S\"  )], columns = S.columns   ))\n",
    "    G4  =  G4.append( pd.DataFrame([[ get_type(frmt), frmt ] + calc(Bytes, part1, part2, part3, frmt, mode=\"G4\" )], columns = G4.columns  ))\n",
    "    G6  =  G6.append( pd.DataFrame([[ get_type(frmt), frmt ] + calc(Bytes, part1, part2, part3, frmt, mode=\"G6\" )], columns = G6.columns  ))\n",
    "    G8  =  G8.append( pd.DataFrame([[ get_type(frmt), frmt ] + calc(Bytes, part1, part2, part3, frmt, mode=\"G8\" )], columns = G8.columns  ))\n",
    "    G4S = G4S.append( pd.DataFrame([[ get_type(frmt), frmt ] + calc(Bytes, part1, part2, part3, frmt, mode=\"G4S\")], columns = G4S.columns ))\n",
    "    G6S = G6S.append( pd.DataFrame([[ get_type(frmt), frmt ] + calc(Bytes, part1, part2, part3, frmt, mode=\"G6S\")], columns = G6S.columns )) \n",
    "    G8S = G8S.append( pd.DataFrame([[ get_type(frmt), frmt ] + calc(Bytes, part1, part2, part3, frmt, mode=\"G8S\")], columns = G8S.columns ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "del SOS, EOI, idats, lists, datas, frames\n",
    "\n",
    "import dill\n",
    "dill.dump_session('dataframes.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 🤖 Training\n",
    "_[> Back to contents](#Contents:)_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Models are built using every algorithm and are trained based on every DataFrame specified. For each case there are two models: one with default settings and one with hyperparameters recommended by `RandomizedSearchCV`. The total number of models is 70 _(7 DataFrames * 5 Algorithms * 2 Sets of hyperparameters)_."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import dill\n",
    "dill.load_session('dataframes.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split     # Splits data into train & test sets\n",
    "from sklearn.metrics import accuracy_score               # Accuracy score\n",
    "from sklearn.model_selection import RandomizedSearchCV   # Randomized search on hyperparameters\n",
    "from sklearn.ensemble import RandomForestClassifier      # Random Forest Classifier\n",
    "from sklearn.neighbors import KNeighborsClassifier       # KNN Classifier\n",
    "from xgboost import XGBClassifier                        # XGBoost Classifier\n",
    "from lightgbm import LGBMClassifier                      # LightGBM Classifier\n",
    "from catboost import CatBoostClassifier                  # CatBoost Classifier\n",
    "from catboost import Pool                                # Dataset processing for CatBoost\n",
    "from lightgbm import log_evaluation                      # \n",
    "from sklearn import preprocessing                        # Encodes target variables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LabelEncoder()"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "leType = preprocessing.LabelEncoder()\n",
    "leType.fit(\"Image Video Audio Executable\".split())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataframes = [\"S\", \"G4\", \"G6\", \"G8\", \"G4S\", \"G6S\", \"G8S\"]\n",
    "\n",
    "for df in dataframes:\n",
    "    \n",
    "    # Creation of 2 sets: data to predict (y) & data to classify with (x)\n",
    "    exec( f\"x{df} = {df}.drop(columns=['Type','Format']).values\" )\n",
    "    exec( f\"y{df} = {df}['Type'].values\" )\n",
    "    exec( f\"y{df} = leType.transform(y{df})\" )\n",
    "    \n",
    "    # Splitting each set into 2 new sets: training & validation\n",
    "    exec( f\"x{df}_train, x{df}_valid, y{df}_train, y{df}_valid = train_test_split(x{df}, y{df}, test_size=0.1, random_state=42)\" )\n",
    "    exec( f\"dataset{df}  = Pool(x{df}_train, y{df}_train)\" )\n",
    "    exec( f\"evalset{df} = [(x{df}_valid, y{df}_valid)]\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random Forest\n",
    "🔗 Link: [Random Forest Classifier](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [00:20<00:00,  2.89s/it]\n"
     ]
    }
   ],
   "source": [
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"rfc{df} = RandomForestClassifier()\" )\n",
    "    exec( f\"rfc{df}.fit(x{df}_train, y{df}_train)\" )\n",
    "\n",
    "    exec( f\"rfcValAcc{df}   = accuracy_score(y{df}_valid, rfc{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"rfcTrainAcc{df} = accuracy_score(y{df}_train, rfc{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Random Forest with recommended hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [06:04<00:00, 52.04s/it]\n"
     ]
    }
   ],
   "source": [
    "RFC = RandomForestClassifier()\n",
    "\n",
    "params={\n",
    "    'max_depth':        [2,5,7,10,15],\n",
    "    'n_estimators':     [25,50,75,100,125,150],\n",
    "    'min_samples_leaf': [1,2,5,7,10]\n",
    "}\n",
    "\n",
    "\n",
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"rfcRec{df} = RandomizedSearchCV(RFC, param_distributions=params)\" )\n",
    "    exec( f\"rfcRec{df}.fit(x{df}_train, y{df}_train)\" )\n",
    "    \n",
    "    exec( f\"rfcRecValAcc{df}   = accuracy_score(y{df}_valid, rfcRec{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"rfcRecTrainAcc{df} = accuracy_score(y{df}_train, rfcRec{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S]\tRecommended params:\t {'n_estimators': 75, 'min_samples_leaf': 2, 'max_depth': 10}\n",
      "[G4]\tRecommended params:\t {'n_estimators': 75, 'min_samples_leaf': 1, 'max_depth': 10}\n",
      "[G6]\tRecommended params:\t {'n_estimators': 150, 'min_samples_leaf': 2, 'max_depth': 15}\n",
      "[G8]\tRecommended params:\t {'n_estimators': 150, 'min_samples_leaf': 1, 'max_depth': 15}\n",
      "[G4S]\tRecommended params:\t {'n_estimators': 50, 'min_samples_leaf': 7, 'max_depth': 15}\n",
      "[G6S]\tRecommended params:\t {'n_estimators': 125, 'min_samples_leaf': 1, 'max_depth': 10}\n",
      "[G8S]\tRecommended params:\t {'n_estimators': 100, 'min_samples_leaf': 2, 'max_depth': 15}\n"
     ]
    }
   ],
   "source": [
    "for df in dataframes:\n",
    "    exec( f\"print('[{df}]\\tRecommended params:\\t', rfcRec{df}.best_params_)\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## K-Nearest Neighbors\n",
    "🔗 Link: [KNN](https://scikit-learn.org/stable/modules/generated/sklearn.neighbors.KNeighborsClassifier.html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [00:18<00:00,  2.68s/it]\n"
     ]
    }
   ],
   "source": [
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"knn{df} = KNeighborsClassifier()\" )\n",
    "    exec( f\"knn{df}.fit(x{df}_train, y{df}_train)\" )\n",
    "\n",
    "    exec( f\"knnValAcc{df}   = accuracy_score(y{df}_valid, knn{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"knnTrainAcc{df} = accuracy_score(y{df}_train, knn{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### K-Nearest Neighbors with recommended hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [02:38<00:00, 22.63s/it]\n"
     ]
    }
   ],
   "source": [
    "KNN = KNeighborsClassifier()\n",
    "\n",
    "params={\n",
    "    'leaf_size':     [10,20,30,40,50],\n",
    "    'n_neighbors':   [3,4,5,6,7,8,9,10]\n",
    "}\n",
    "\n",
    "\n",
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"knnRec{df} = RandomizedSearchCV(KNN, param_distributions=params)\" )\n",
    "    exec( f\"knnRec{df}.fit(x{df}_train, y{df}_train)\" )\n",
    "    \n",
    "    exec( f\"knnRecValAcc{df}   = accuracy_score(y{df}_valid, knnRec{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"knnRecTrainAcc{df} = accuracy_score(y{df}_train, knnRec{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S]\tRecommended params:\t {'n_neighbors': 5, 'leaf_size': 40}\n",
      "[G4]\tRecommended params:\t {'n_neighbors': 3, 'leaf_size': 40}\n",
      "[G6]\tRecommended params:\t {'n_neighbors': 3, 'leaf_size': 40}\n",
      "[G8]\tRecommended params:\t {'n_neighbors': 3, 'leaf_size': 20}\n",
      "[G4S]\tRecommended params:\t {'n_neighbors': 4, 'leaf_size': 10}\n",
      "[G6S]\tRecommended params:\t {'n_neighbors': 3, 'leaf_size': 10}\n",
      "[G8S]\tRecommended params:\t {'n_neighbors': 5, 'leaf_size': 50}\n"
     ]
    }
   ],
   "source": [
    "for df in dataframes:\n",
    "    exec( f\"print('[{df}]\\tRecommended params:\\t', knnRec{df}.best_params_)\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## XGBoost\n",
    "🔗 Link: [XGBoost](https://github.com/dmlc/xgboost/)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [00:20<00:00,  2.99s/it]\n"
     ]
    }
   ],
   "source": [
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"xgb{df} = XGBClassifier(tree_method='gpu_hist', predictor='gpu_predictor', eval_metric='mlogloss')\" )\n",
    "    exec( f\"xgb{df}.fit(x{df}_train, y{df}_train, eval_set=evalset{df}, verbose=False)\" )\n",
    "\n",
    "    exec( f\"xgbValAcc{df}   = accuracy_score(y{df}_valid, xgb{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"xgbTrainAcc{df} = accuracy_score(y{df}_train, xgb{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### XGBoost with recommended hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [1:26:24<00:00, 740.68s/it]\n"
     ]
    }
   ],
   "source": [
    "XGB = XGBClassifier(tree_method='gpu_hist', predictor='gpu_predictor', eval_metric='mlogloss')\n",
    "\n",
    "params={\n",
    "    'max_depth':        [2,5,7,10],\n",
    "    'max_leaves':       [16,32,64,96],\n",
    "    'n_estimators':     [100,250,500,750,1000],\n",
    "    'learning_rate':    [0.01,0.025,0.05,0.075,0.1]\n",
    "}\n",
    "\n",
    "\n",
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"xgbRec{df} = RandomizedSearchCV(XGB, param_distributions=params)\" )\n",
    "    exec( f\"xgbRec{df}.fit(x{df}_train, y{df}_train, eval_set=evalset{df}, verbose=False)\" )\n",
    "    \n",
    "    exec( f\"xgbRecValAcc{df}   = accuracy_score(y{df}_valid, xgbRec{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"xgbRecTrainAcc{df} = accuracy_score(y{df}_train, xgbRec{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S]\tRecommended params:\t {'n_estimators': 750, 'max_leaves': 32, 'max_depth': 10, 'learning_rate': 0.075}\n",
      "[G4]\tRecommended params:\t {'n_estimators': 500, 'max_leaves': 16, 'max_depth': 10, 'learning_rate': 0.1}\n",
      "[G6]\tRecommended params:\t {'n_estimators': 500, 'max_leaves': 32, 'max_depth': 7, 'learning_rate': 0.1}\n",
      "[G8]\tRecommended params:\t {'n_estimators': 750, 'max_leaves': 32, 'max_depth': 5, 'learning_rate': 0.1}\n",
      "[G4S]\tRecommended params:\t {'n_estimators': 750, 'max_leaves': 32, 'max_depth': 7, 'learning_rate': 0.075}\n",
      "[G6S]\tRecommended params:\t {'n_estimators': 500, 'max_leaves': 16, 'max_depth': 5, 'learning_rate': 0.075}\n",
      "[G8S]\tRecommended params:\t {'n_estimators': 750, 'max_leaves': 16, 'max_depth': 7, 'learning_rate': 0.1}\n"
     ]
    }
   ],
   "source": [
    "for df in dataframes:\n",
    "    exec( f\"print('[{df}]\\tRecommended params:\\t', xgbRec{df}.best_params_)\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LightGBM\n",
    "🔗 Link: [LightGBM](https://github.com/Microsoft/LightGBM/)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [00:12<00:00,  1.73s/it]\n"
     ]
    }
   ],
   "source": [
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"lgbm{df} = LGBMClassifier()\" )\n",
    "    exec( f\"lgbm{df}.fit(x{df}_train, y{df}_train, eval_metric='logloss', eval_set=evalset{df}, callbacks=[log_evaluation(period=0)])\" )\n",
    "\n",
    "    exec( f\"lgbmValAcc{df}   = accuracy_score(y{df}_valid, lgbm{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"lgbmTrainAcc{df} = accuracy_score(y{df}_train, lgbm{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### LGBM with recommended hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [23:53<00:00, 204.80s/it]\n"
     ]
    }
   ],
   "source": [
    "LGBM = LGBMClassifier()\n",
    "\n",
    "params={\n",
    "    'max_depth':        [2,5,7,10],\n",
    "    'num_leaves':       [16,32,64,96],\n",
    "    'n_estimators':     [100,250,500,750,1000],\n",
    "    'learning_rate':    [0.01,0.025,0.05,0.075,0.1]\n",
    "}\n",
    "\n",
    "\n",
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"lgbmRec{df} = RandomizedSearchCV(LGBM, param_distributions=params)\" )\n",
    "    exec( f\"lgbmRec{df}.fit(x{df}_train, y{df}_train, eval_set=evalset{df}, callbacks=[log_evaluation(period=0)])\" )\n",
    "    \n",
    "    exec( f\"lgbmRecValAcc{df}   = accuracy_score(y{df}_valid, lgbmRec{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"lgbmRecTrainAcc{df} = accuracy_score(y{df}_train, lgbmRec{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S]\tRecommended params:\t {'num_leaves': 16, 'n_estimators': 250, 'max_depth': 10, 'learning_rate': 0.075}\n",
      "[G4]\tRecommended params:\t {'num_leaves': 16, 'n_estimators': 750, 'max_depth': 10, 'learning_rate': 0.1}\n",
      "[G6]\tRecommended params:\t {'num_leaves': 16, 'n_estimators': 500, 'max_depth': 5, 'learning_rate': 0.1}\n",
      "[G8]\tRecommended params:\t {'num_leaves': 16, 'n_estimators': 1000, 'max_depth': 5, 'learning_rate': 0.075}\n",
      "[G4S]\tRecommended params:\t {'num_leaves': 64, 'n_estimators': 1000, 'max_depth': 5, 'learning_rate': 0.075}\n",
      "[G6S]\tRecommended params:\t {'num_leaves': 16, 'n_estimators': 250, 'max_depth': 10, 'learning_rate': 0.1}\n",
      "[G8S]\tRecommended params:\t {'num_leaves': 32, 'n_estimators': 1000, 'max_depth': 10, 'learning_rate': 0.075}\n"
     ]
    }
   ],
   "source": [
    "for df in dataframes:\n",
    "    exec( f\"print('[{df}]\\tRecommended params:\\t', lgbmRec{df}.best_params_)\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CatBoost\n",
    "🔗 Link: [CatBoost](https://github.com/catboost/catboost/)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [01:00<00:00,  8.65s/it]\n"
     ]
    }
   ],
   "source": [
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"cbc{df} = CatBoostClassifier(task_type='GPU')\" )\n",
    "    exec( f\"cbc{df}.fit(x{df}_train, y{df}_train, use_best_model=True, eval_set=dataset{df}, verbose=False)\" )\n",
    "\n",
    "    exec( f\"cbcValAcc{df}   = accuracy_score(y{df}_valid, cbc{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"cbcTrainAcc{df} = accuracy_score(y{df}_train, cbc{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### CatBoost with recommended hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "DataFrame: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 7/7 [42:44<00:00, 366.40s/it]\n"
     ]
    }
   ],
   "source": [
    "CBC = CatBoostClassifier(task_type='GPU')\n",
    "\n",
    "params={\n",
    "    'max_depth':        [2,5,7,10],\n",
    "    'n_estimators':     [100,250,500,750,1000],\n",
    "    'learning_rate':    [0.01,0.025,0.05,0.075,0.1]\n",
    "}\n",
    "\n",
    "\n",
    "for df in tqdm(dataframes, desc='DataFrame'):\n",
    "    exec( f\"cbcRec{df} = RandomizedSearchCV(CBC, param_distributions=params)\" )\n",
    "    exec( f\"cbcRec{df}.fit(x{df}_train, y{df}_train, use_best_model=True, eval_set=evalset{df}, verbose=False)\" )\n",
    "    \n",
    "    exec( f\"cbcRecValAcc{df}   = accuracy_score(y{df}_valid, cbcRec{df}.predict(x{df}_valid))\" )\n",
    "    exec( f\"cbcRecTrainAcc{df} = accuracy_score(y{df}_train, cbcRec{df}.predict(x{df}_train))\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S]\tRecommended params:\t {'n_estimators': 1000, 'max_depth': 7, 'learning_rate': 0.1}\n",
      "[G4]\tRecommended params:\t {'n_estimators': 500, 'max_depth': 7, 'learning_rate': 0.1}\n",
      "[G6]\tRecommended params:\t {'n_estimators': 750, 'max_depth': 5, 'learning_rate': 0.1}\n",
      "[G8]\tRecommended params:\t {'n_estimators': 1000, 'max_depth': 5, 'learning_rate': 0.1}\n",
      "[G4S]\tRecommended params:\t {'n_estimators': 1000, 'max_depth': 2, 'learning_rate': 0.05}\n",
      "[G6S]\tRecommended params:\t {'n_estimators': 750, 'max_depth': 5, 'learning_rate': 0.1}\n",
      "[G8S]\tRecommended params:\t {'n_estimators': 500, 'max_depth': 7, 'learning_rate': 0.075}\n"
     ]
    }
   ],
   "source": [
    "for df in dataframes:\n",
    "    exec( f\"print('[{df}]\\tRecommended params:\\t', cbcRec{df}.best_params_)\" )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "del datasetS, datasetG4, datasetG6, datasetG8, datasetG4S, datasetG6S, datasetG8S\n",
    "del evalsetS, evalsetG4, evalsetG6, evalsetG8, evalsetG4S, evalsetG6S, evalsetG8S\n",
    "\n",
    "import dill\n",
    "dill.dump_session('models.pkl')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 📊 Comparison\n",
    "_[> Back to contents](#Contents:)_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All models trained based on different DataFrames are grouped depending on the algorithm used and are compared with each other. The models with default settings are compared with the ones with recommended hyperparameters. Every model's accuracy of predictions on validation data and accuracy of predictions on train data are also compared with each other."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "import seaborn as sn\n",
    "from numpy import newaxis\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<svg  width=\"330\" height=\"55\"><rect x=\"0\" y=\"0\" width=\"55\" height=\"55\" style=\"fill:#f5efe6;stroke-width:2;stroke:rgb(255,255,255)\"/><rect x=\"55\" y=\"0\" width=\"55\" height=\"55\" style=\"fill:#f6dee3;stroke-width:2;stroke:rgb(255,255,255)\"/><rect x=\"110\" y=\"0\" width=\"55\" height=\"55\" style=\"fill:#e9d2ee;stroke-width:2;stroke:rgb(255,255,255)\"/><rect x=\"165\" y=\"0\" width=\"55\" height=\"55\" style=\"fill:#c6d1ed;stroke-width:2;stroke:rgb(255,255,255)\"/><rect x=\"220\" y=\"0\" width=\"55\" height=\"55\" style=\"fill:#a8d5cf;stroke-width:2;stroke:rgb(255,255,255)\"/><rect x=\"275\" y=\"0\" width=\"55\" height=\"55\" style=\"fill:#a5cea3;stroke-width:2;stroke:rgb(255,255,255)\"/></svg>"
      ],
      "text/plain": [
       "[(0.9606673315011522, 0.9386660821471728, 0.9014843466101898),\n",
       " (0.9650791968125461, 0.8718625177766127, 0.8912563656651196),\n",
       " (0.9144322710296299, 0.8235878418953423, 0.931806264970016),\n",
       " (0.775683509263619, 0.8214698728227631, 0.9294144419726944),\n",
       " (0.6592653869469826, 0.8359820945603558, 0.8125726947429762),\n",
       " (0.648385100679158, 0.8075016938524312, 0.6384979693454241)]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "params = 'ch:start=2,rot=4,dark=0.74,light=0.94'\n",
    "palette = sn.color_palette(params, as_cmap=True)\n",
    "sn.color_palette(params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "types = [\"IMG\",\"VID\",\"AUD\",\"EXE\"]\n",
    "\n",
    "def compareHeatmaps(model, modelRec, X, Y):\n",
    "    \n",
    "    prediction     = model.predict(X)\n",
    "    predictionRec = modelRec.predict(X)\n",
    "    \n",
    "    matrix     = confusion_matrix(Y, prediction)\n",
    "    matrixRec = confusion_matrix(Y, predictionRec)\n",
    "    matrix     = pd.DataFrame(matrix, index=types, columns=types)\n",
    "    matrixRec = pd.DataFrame(matrixRec, index=types, columns=types)\n",
    "    matrix     = matrix.astype('float') / matrix.sum(axis=1)\n",
    "    matrixRec = matrixRec.astype('float') / matrixRec.sum(axis=1)\n",
    "    \n",
    "    fig, (ax,axRec) = plt.subplots(ncols=2)\n",
    "    fig.subplots_adjust(wspace=0.042)\n",
    "    \n",
    "    sn.heatmap(matrix, cmap=palette, annot=True, fmt='.0%', linewidths=0.5, ax=ax, cbar=False)\n",
    "    sn.heatmap(matrixRec, cmap=palette, annot=True, fmt='.0%', linewidths=0.5, ax=axRec, cbar=False)\n",
    "    \n",
    "\n",
    "    ax.tick_params(rotation=0)\n",
    "    axRec.tick_params(rotation=0)\n",
    "    axRec.yaxis.tick_right()\n",
    "    \n",
    "    \n",
    "    fig.text(0.415, 0.000, \"Guessed type\", size=12, weight='bold')\n",
    "    fig.text(0.010, 0.415, \"True type\",    size=12, weight='bold', rotation='vertical')\n",
    "    ax.set_title(\"Default params\",   size=14, pad=8)\n",
    "    axRec.set_title(\"Recommended params\", size=14, pad=8)\n",
    "    \n",
    "    \n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def showAccuracy(alg):\n",
    "    for df in dataframes:\n",
    "        exec( f\"print('[{df}] Validation accuracy:', {alg}ValAcc{df}  )\" )\n",
    "        exec( f\"print('[{df}] Train accuracy:',      {alg}TrainAcc{df})\" )\n",
    "        exec( f\"print('[{df}] [Recommended] Validation accuracy:', {alg}RecValAcc{df}  )\" )\n",
    "        exec( f\"print('[{df}] [Recommended] Train accuracy:',      {alg}RecTrainAcc{df})\" )\n",
    "        print('\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S] Validation accuracy: 0.9291666666666667\n",
      "[S] Train accuracy: 0.9999074074074074\n",
      "[S] [Recommended] Validation accuracy: 0.895\n",
      "[S] [Recommended] Train accuracy: 0.9481481481481482\n",
      "\n",
      "\n",
      "[G4] Validation accuracy: 0.9341666666666667\n",
      "[G4] Train accuracy: 0.9944444444444445\n",
      "[G4] [Recommended] Validation accuracy: 0.9116666666666666\n",
      "[G4] [Recommended] Train accuracy: 0.9371296296296296\n",
      "\n",
      "\n",
      "[G6] Validation accuracy: 0.9141666666666667\n",
      "[G6] Train accuracy: 0.9940740740740741\n",
      "[G6] [Recommended] Validation accuracy: 0.9191666666666667\n",
      "[G6] [Recommended] Train accuracy: 0.962037037037037\n",
      "\n",
      "\n",
      "[G8] Validation accuracy: 0.9208333333333333\n",
      "[G8] Train accuracy: 0.9929629629629629\n",
      "[G8] [Recommended] Validation accuracy: 0.925\n",
      "[G8] [Recommended] Train accuracy: 0.9726851851851852\n",
      "\n",
      "\n",
      "[G4S] Validation accuracy: 0.9425\n",
      "[G4S] Train accuracy: 0.999537037037037\n",
      "[G4S] [Recommended] Validation accuracy: 0.9375\n",
      "[G4S] [Recommended] Train accuracy: 0.9594444444444444\n",
      "\n",
      "\n",
      "[G6S] Validation accuracy: 0.9441666666666667\n",
      "[G6S] Train accuracy: 0.9993518518518518\n",
      "[G6S] [Recommended] Validation accuracy: 0.905\n",
      "[G6S] [Recommended] Train accuracy: 0.9296296296296296\n",
      "\n",
      "\n",
      "[G8S] Validation accuracy: 0.9308333333333333\n",
      "[G8S] Train accuracy: 0.9993518518518518\n",
      "[G8S] [Recommended] Validation accuracy: 0.9233333333333333\n",
      "[G8S] [Recommended] Train accuracy: 0.9746296296296296\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "showAccuracy('rfc')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compareHeatmaps(rfcG8, rfcRecG8, xG8_valid, yG8_valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## K-Nearest Neighbors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S] Validation accuracy: 0.7641666666666667\n",
      "[S] Train accuracy: 0.8510185185185185\n",
      "[S] [Recommended] Validation accuracy: 0.7641666666666667\n",
      "[S] [Recommended] Train accuracy: 0.8510185185185185\n",
      "\n",
      "\n",
      "[G4] Validation accuracy: 0.8008333333333333\n",
      "[G4] Train accuracy: 0.8517592592592592\n",
      "[G4] [Recommended] Validation accuracy: 0.8008333333333333\n",
      "[G4] [Recommended] Train accuracy: 0.8517592592592592\n",
      "\n",
      "\n",
      "[G6] Validation accuracy: 0.7866666666666666\n",
      "[G6] Train accuracy: 0.8543518518518518\n",
      "[G6] [Recommended] Validation accuracy: 0.8083333333333333\n",
      "[G6] [Recommended] Train accuracy: 0.894537037037037\n",
      "\n",
      "\n",
      "[G8] Validation accuracy: 0.7683333333333333\n",
      "[G8] Train accuracy: 0.8298148148148148\n",
      "[G8] [Recommended] Validation accuracy: 0.7816666666666666\n",
      "[G8] [Recommended] Train accuracy: 0.8655555555555555\n",
      "\n",
      "\n",
      "[G4S] Validation accuracy: 0.745\n",
      "[G4S] Train accuracy: 0.8184259259259259\n",
      "[G4S] [Recommended] Validation accuracy: 0.75\n",
      "[G4S] [Recommended] Train accuracy: 0.8629629629629629\n",
      "\n",
      "\n",
      "[G6S] Validation accuracy: 0.7975\n",
      "[G6S] Train accuracy: 0.855\n",
      "[G6S] [Recommended] Validation accuracy: 0.8066666666666666\n",
      "[G6S] [Recommended] Train accuracy: 0.8880555555555556\n",
      "\n",
      "\n",
      "[G8S] Validation accuracy: 0.81\n",
      "[G8S] Train accuracy: 0.8429629629629629\n",
      "[G8S] [Recommended] Validation accuracy: 0.8075\n",
      "[G8S] [Recommended] Train accuracy: 0.8824074074074074\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "showAccuracy('knn')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compareHeatmaps(knnG8, knnRecG8, xG8_valid, yG8_valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S] Validation accuracy: 0.9608333333333333\n",
      "[S] Train accuracy: 0.9999074074074074\n",
      "[S] [Recommended] Validation accuracy: 0.9625\n",
      "[S] [Recommended] Train accuracy: 0.9999074074074074\n",
      "\n",
      "\n",
      "[G4] Validation accuracy: 0.9433333333333334\n",
      "[G4] Train accuracy: 0.9930555555555556\n",
      "[G4] [Recommended] Validation accuracy: 0.9525\n",
      "[G4] [Recommended] Train accuracy: 0.992037037037037\n",
      "\n",
      "\n",
      "[G6] Validation accuracy: 0.9166666666666666\n",
      "[G6] Train accuracy: 0.9909259259259259\n",
      "[G6] [Recommended] Validation accuracy: 0.9266666666666666\n",
      "[G6] [Recommended] Train accuracy: 0.9925925925925926\n",
      "\n",
      "\n",
      "[G8] Validation accuracy: 0.9425\n",
      "[G8] Train accuracy: 0.9912962962962963\n",
      "[G8] [Recommended] Validation accuracy: 0.9433333333333334\n",
      "[G8] [Recommended] Train accuracy: 0.9923148148148148\n",
      "\n",
      "\n",
      "[G4S] Validation accuracy: 0.9641666666666666\n",
      "[G4S] Train accuracy: 0.9994444444444445\n",
      "[G4S] [Recommended] Validation accuracy: 0.9658333333333333\n",
      "[G4S] [Recommended] Train accuracy: 0.999537037037037\n",
      "\n",
      "\n",
      "[G6S] Validation accuracy: 0.9541666666666667\n",
      "[G6S] Train accuracy: 0.9993518518518518\n",
      "[G6S] [Recommended] Validation accuracy: 0.9516666666666667\n",
      "[G6S] [Recommended] Train accuracy: 0.9964814814814815\n",
      "\n",
      "\n",
      "[G8S] Validation accuracy: 0.9483333333333334\n",
      "[G8S] Train accuracy: 0.9992592592592593\n",
      "[G8S] [Recommended] Validation accuracy: 0.9525\n",
      "[G8S] [Recommended] Train accuracy: 0.9993518518518518\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "showAccuracy('xgb')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAnUAAAHYCAYAAAArnR75AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAA9hAAAPYQGoP6dpAABtqElEQVR4nO3dd1RURxsG8GfpvYogFrrYe4tGwY4FMRo1CkYFe03UqGjsxm7sGitojD323qJi7w0RG2BDjSAoVWDn+4OPjeuCAlIvz++cPbBzZ2ff2Xtn9925ZWVCCAEiIiIiKtTU8jsAIiIiIvp6TOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqKN+8e/cOw4YNg52dHTQ1NSGTyXDjxo1cf97Q0FDIZDL07Nkz15+LiKTJ1dUVMpks19r39/eHTCaDv79/rj0HSQ+TOlJIS3Y+vunp6cHa2hpNmzbFhAkT8OjRoxx7vlGjRmHRokWoVKkSxowZg4kTJ8LKyirH2s+q3H6TpqIrvbGlqamJkiVLonPnzrhy5Up+h0hEEqCR3wFQwePg4AAvLy8AQGJiIl6/fo1Lly5h6tSpmD59OkaNGoXffvvtqxOgffv2oWzZsti7d29OhE1U4H08tmJjY3H16lVs27YNu3btwrFjx9CoUaN8jpCICjMmdaTC0dERkyZNUik/c+YMunfvjhkzZkBdXR1Tp079qud58eIFP8SoSElvbM2cORO+vr4YP348Tp06lT+BEZEkcPcrZdq3336LQ4cOQVtbG7Nnz8bTp09V6uzevRtNmzaFqakpdHR0UKlSJcydOxcpKSmKOj179oRMJoMQAqdOnVLsjnJ1dQUAREdHY9asWXBxcYG1tTW0tLRgbW2NH3/8Md3dv2nthYaGqiybNGkSZDIZTp48+dm+yWQyxQfqx7vIMnPcna2tLWxtbREVFYV+/frBysoKOjo6qF69OjZt2qRS/8WLF5g4cSLq1auH4sWLQ1tbG7a2thg4cCBev36dYf8eP36MefPmoUKFCtDW1lbE9jXtzZ07F2XLloWuri4qVKiAzZs3AwA+fPiAcePGwdbWFjo6OqhSpQoOHjyo0lZ4eDiGDRsGJycn6OrqwsTEBOXLl0f//v0RHR39xdeOAB8fHwDA1atXVZZ9+PABv//+O2rUqAF9fX0YGhqiYcOG2LNnT7ptffjwAfPnz0ft2rVhaGgIAwMDVKhQAcOHD8fbt2+V6t65cwedO3dWbDN2dnb46aefEBERodJu2jYeHR2NAQMGoESJEtDX10ejRo1w7do1AKnboZeXF4oXLw5dXV20aNECDx48UGkrbaw/f/4c3bp1Q7FixWBoaIg2bdrg8ePHAICgoCC0b98eZmZmMDQ0xPfff49Xr16l2+dbt27hhx9+QIkSJaClpQUbGxsMGTJEpR8fH0v78OFDfPfddzA1NYW+vj6aNWuGmzdvptv+mTNn4OLiAn19fZibm6NLly7pvvelEUJg7dq1aNCgAYyMjKCnp4datWph7dq16daPjIxE//79YWlpCT09PdSuXRs7d+7MsP2M8H2I70MAZ+ooi5ydndG5c2f8+eef2LVrF4YMGaJY5uvri5kzZ6JkyZLo0KEDjI2NERAQgF9++QUXL17Etm3bAADt27eHra0tJk+eDBsbG8Wbgq2tLYDUN/QJEyagcePG+O6776Cvr4979+5h48aN2L9/P65duwYbG5sc7dfEiRPh7++PsLAwTJw4UVFerVq1TD3+w4cPaNasGWJiYtC9e3fExsZi69at6NatG968eaP0Op0+fRrz5s1D06ZNUbduXWhqauL69etYvnw5Dh8+jGvXrsHY2FjlOYYMGYILFy6gTZs2cHd3R/Hixb+qveHDh+PixYtwd3eHuro6Nm/ejG7dusHU1BSLFy/G3bt30aZNGyQkJGDjxo3w8PBAUFAQHBwcAABxcXFo0KABQkND0aJFC3z33Xf48OEDQkJC8Oeff2LkyJHpPi+lT0ND+e04MTERbm5uOHnyJKpVqwYfHx8kJSVh//798PDwwOLFizF48GBF/fj4eDRv3hxnz56Fk5MTevXqBW1tbTx48AArVqzAjz/+CFNTUwCpiUrLli3x4cMHfP/997C1tcX58+excOFC7Nu3DxcuXECxYsWU4vnw4QOaN2+OhIQEdOnSBa9evcLWrVvRrFkznDt3Di1btkSJEiXg5eWFhw8fYu/evWjTpg2CgoKgrq6u1Nbbt2/x7bffwsrKCj169MD9+/exb98+3Lt3D7t370bDhg1Rs2ZNeHt74+rVq/j7778RGRmJEydOKLWzZ88edO7cGWpqavDw8EDp0qVx9+5dLFmyBIcPH8bFixcVfU4TGhqKevXqoWLFivD29sajR4+we/duNG7cGEFBQbC0tFTUPX78OFq1agU1NTV06dIF1tbWOH78OBo0aKDSLpCa0Hl6emLTpk1wcnJCt27doKWlhaNHj8LHxwd3797F3LlzFfXj4uLg6uqK27dv45tvvoGLiwuePn2KLl26oEWLFpnZbFTWEd+Hivj7kCD6v5CQEAFAtGzZ8rP11qxZIwCI7t27K8qOHDmieGxMTIyiXC6Xi/79+wsAYvv27UrtABAuLi4q7UdFRYmIiAiV8hMnTgg1NTXRu3dvpfIePXoIACIkJETlMRMnThQAxD///KPSzx49eijVdXFxEdkZEjY2NgKAaNSokUhMTFSUP336VBQrVkxoa2uLZ8+eKcpfvXol3r9/r9LOunXrBAAxbdq0dPtXqlQpERYWpvK47LZXtmxZ8fr1a0X5xYsXBQBhYmIivv32W6X1uGXLFgFADBkyRFG2Z88eAUD89NNPKs/9/v17kZCQoFJeVH1ubE2fPl0AEG3atFEqHzt2rAAgxo8fL+RyuaL83bt3olatWkJLS0s8f/5cUT5ixAjFuExOTlZqKyoqSrGNpKSkCAcHBwFAHDp0SKneL7/8IgAIb29vpfK0bbxTp04iKSlJUT5r1izFNvPzzz8rxTlgwAABQPz9999KbQEQAMTPP/+sVJ5W38TERCxYsEBRLpfLRevWrQUAcfXqVUX5mzdvhJGRkShZsqQIDQ1VamvTpk0CgBg8eLCiLG0dABAzZ85Uqv/rr78KAGLGjBmKspSUFGFvby9kMpkICAhQiqdbt26Ktj62cuVKAUD06tVLfPjwQVGemJgo3N3dBQBx5coVRXna+1OfPn2U2jl06JCifT8/P5EZfB/i+5AQQjCpI4XMJnUHDx4UAESrVq0UZe3atRMA0h3sUVFRQiaTiY4dOyqVZ5TUfU7lypWFra2tUllBSerOnDmjsmzq1KkCgJg7d+4X25HL5cLIyEi4uroqlaf1b+HChVmK60vtrVu3TuUx9vb2AoA4deqUUnlycrLQ1NQUjRo1UpSlvZn6+vpmKa6iKG2bc3BwEBMnThQTJ04UI0eOFI0bNxYAhKWlpbh7966ifkpKijA1NRUODg5KiVKatNd+8eLFQgghkpKShKGhoTA2NhaRkZGfjeX06dMq4zfN+/fvhZmZmdDR0VFKDNK28U/H95MnTwQAYWBgIGJjY9N9ngkTJiiVf6l+en1ev369ACDWrl2rKPv9998FALF+/fp0+1mjRg1RrFgxxf20dWBnZydSUlKU6qYt69Chg6Ls1KlTAoBwd3dXaTs0NFSoq6urvF9UqVJF6Ovri7i4OJXH3Lp1SwAQI0aMUJTZ2dkJLS0tER4erlK/adOm2Urq+D5UtHH3K+WICxcuQF9fP8PjRnR1dXHv3r1Mt3fy5EksWLAAFy9exJs3b5CcnKxYpqWl9dXx5jQNDQ188803KuUNGzYEAFy/fl2pfMeOHVixYgWuXbuGt2/fKh1z+OLFi3Sfo06dOhk+f3baS2/XcokSJfD48WOVZerq6ihevLhSW40aNUKJEiUwc+ZM3Lx5E23btoWLiwvKly/PS8Nk4NGjR5g8ebJSmZWVFQICAuDo6KgoCw4Oxtu3b2Ftba1SHwD+/fdfAFCMqXv37uH9+/do1qxZursFP5a2LaYdw/oxAwMD1KpVC0eOHEFwcDAqV66sWGZqaooyZcoo1S9RogQAwMnJCXp6eukuS2/7+1z9KlWqqGw/6bV14cIFAMDFixfTPdY2ISEBb968wZs3b5R2JVerVg1qasqHk5cqVQoAEBUVpShLO8YubQx/zMbGBqVLl1Y6jjcuLg63b9+GtbU1Zs2apfKYpKQkAP+ts3fv3iEkJAQVKlRI91JODRs2xPHjx1XKP4fvQ3wfYlJHWZY2oCwsLBRlkZGRSE5OTvcDKE1sbGym2t+2bRu6dOkCAwMDtGzZEra2ttDT01NciDMsLOzrOpALihUrpvJBAUBxfM7HB+vOmzcPI0eOhIWFBVq0aIFSpUpBV1cXALBgwQIkJiam+xwfH+vzsey2Z2RkpFKWdlxXRsvSPpgAwNjYGBcuXMCECROwd+9eHDhwAABQunRpjBkzBgMHDkz3eYuyli1b4tChQwBSE7N169Zh9OjRaNeuHS5dugQDAwMAqeMJAAIDAxEYGJhhe2ljKm37Klmy5BdjePfuHYCMt6e0BCqtXprsbC8AlLaZnGwr7TVaunRpOr34T2xsrFJS97n2P05C0l7TtGPGPmVpaamU1L19+xZCCDx//jxT74Npr+/n2s8qvg/xfYhJHWVZ2pmktWvXVpQZGRlBJpPhzZs3X93+pEmToKOjg6tXr8LJyUlpWdpZUR9LexP7eDYvTV6d+fTmzRvI5XKVN9S0M/bSDtRNTk7G1KlTUaJECdy4cUPpDV0IgdmzZ2f4HOl96/ya9nJCmTJl4O/vD7lcjlu3buHIkSNYtGgRBg0aBFNTU3Tt2jVXn78ws7CwwMiRIxEdHY1p06bh119/xYIFCwD892HWsWNHbN++/YttmZiYAACeP3/+xbppbWd0NunLly+V6hVUafHdvn0blSpVyvH208ZsemduAqqvX1o8NWvWzNTFpNPqZ7b9zOD7EN+HeEkTypL79+9j69at0NbWxnfffacor1u3LiIiItK9hEFWPXr0COXLl1dJ6MLDwxWXPPhY2u6m9D7QPt3d8DlpZ+h9/G09s5KTk3H+/HmV8oCAAABA9erVAaS+6UZHR+Obb75R+YZ+5coVxMfHZ+l5c7q97FJTU0O1atUwatQoxeUTMrrsBikbO3YsrK2tsWzZMsXMT/ny5WFkZIQrV66kO9P1KWdnZxgZGeHy5csqly75VNq2mN5lfmJjY3HlyhXo6urC2dk5y33JS3Xr1gWAdMddTqhatSqA/8bwx8LCwlQua2JoaIjy5csjKChIaTduRoyMjGBnZ4eHDx8qEumPpfe8X8L3Ib4PMamjTDt79ixatmyJxMREjBkzRmlXz9ChQwEA3t7e6V7n6uXLlwgKCsrU89jY2ODhw4dK31QTEhIwYMCAdD/g0mYMP/2NxO3bt2fpYq5mZmYA8NlrUH3O2LFj8eHDB8X9Z8+eYeHChdDW1sYPP/wAAIpreF27dg1xcXGKum/fvlW63EBm5XR7WREYGJjubEJamY6OTq4+v1To6upi9OjRSEpKUlzQW0NDAwMGDEBYWBhGjhyZ7nZ/584dxSyPhoYG+vXrh+joaAwbNkzli0l0dDRiYmIAAA0aNICDgwMOHjyIY8eOKdWbNm0aIiIi0LVr1wJ57OrHevXqBUNDQ4wbNy7dXdRxcXGK4+6y49tvv4WdnR327duHM2fOKMqFEBg7dmy6X/6GDh2KuLg49OnTJ93DTUJCQpR22Xbv3h0fPnzAhAkTlOodOXIky8fTpeH7UKqi+j7E3a+k4uHDh4qr3n/48EHxM2G3b9+Guro6fv31V6VruQGAm5sbxo8fj6lTp8LR0RFubm6wsbFBREQEHj58iICAAEybNg3ly5f/4vMPGTIEQ4YMQfXq1fH9998jOTkZR48ehRACVatWVblIqIeHBxwcHODv74+nT5+ievXqCAoKwokTJ9C6dWvFMRZf0qRJE2zfvh0dO3ZEq1atoKOjg6pVq8Ld3f2Ljy1RogRiY2NRpUoVuLu7K64PFRERgUWLFikSYDU1NQwcOBDz5s1TtP3u3TscPHgQNjY2sLa2zlSsaXK6vaw4evQofvnlFzRo0ABly5aFubk5Hj9+jD179kBHRweDBg3KteeWmr59+2LWrFlYv349xo4dCwcHB0yePBnXrl3DokWLsH//fjRq1AjFixfH8+fPcfv2bdy8eRPnz59XzIxMmTIFFy5cwJ9//okLFy6gVatW0NbWxuPHj3Ho0CGcOXNGcZKAv78/WrZsidatW6NTp06wsbHB+fPncfLkSTg4OGDmzJn5/Ip8mYWFBTZt2oROnTqhatWqcHNzQ7ly5ZCYmIjQ0FCcOnUK9evXVxzDmFVqampYuXIlWrdujWbNmimuU3fixAmEh4ejSpUquHXrltJj+vXrhwsXLmDdunU4e/YsmjVrBmtra7x69Qr37t3DxYsXsXHjRsU1OUeNGoUdO3Zg1apVCAwMRKNGjfD06VNs3boVbdq0wf79+7MUM9+H+D7ES5qQwsfXcUq76erqihIlSojGjRuL8ePHi4cPH362jaNHjwp3d3dhYWEhNDU1hZWVlfjmm2/E1KlTxZMnT5TqIoNLmsjlcvHHH3+IihUrCh0dHWFlZSV8fHzE69evM7zsSEhIiGjfvr0wNDQU+vr6omnTpuLy5ctZuqRJUlKSGDVqlChTpozQ0NBIt056bGxshI2NjYiMjBR9+/YVlpaWQltbW1StWlVs3LhRpf6HDx/Eb7/9JpycnIS2trYoU6aMGDFihHj//r2irY997pItOd3e5y7r8mlbd+/eFcOGDRPVq1cX5ubmQltbW9jb24sePXqIwMDAz71kRU5mLhe0ePFiles/JicnixUrVogGDRoIIyMjxfp1c3MTy5cvV7qGlxBCJCQkiLlz54pq1aoJXV1dYWBgICpUqCBGjBgh3r59q1T31q1b4vvvvxfFihUTmpqawsbGRgwbNkz8+++/KrGltx2lyWgcZzTOslpfCCH++ecfAUBMnDhRZdm9e/eEj4+PsLGxEVpaWsLU1FRUrlxZDB06VFy6dClT7X8urtOnT4tGjRoJXV1dYWZmJjp16iTCwsI+O1a2bNkimjVrJkxNTYWmpqYoWbKkcHV1FfPmzVN5fSMiIkTfvn2FhYWF0NHRETVr1hQ7duwQfn5+Wb6kCd+H+D4kE0KIvEogiaQo7Vt3ej9TRkSUF/g+RACPqSMiIiKSBCZ1RERERBLApI6IiIhIAnhMHREREZEEcKaOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAjfwOQIo2BJ7I7xDyjFfFJoiLeJ7fYeQpPfOSiP/3WX6Hkad0LUohLio8v8PIM3omJbJUn2Ne2jjmi4asjvuCiDN1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCRAEkldz5490b59e8X/MpkM/fv3V6k3aNAgyGQy9OzZU6n85cuXGDZsGBwdHaGjowNLS0s0aNAAy5cvR1xcXB704D/yFDn+2bgHi/v/ihk/DMWSAeNxeusBCCEUdaZ2GJDu7dyuIwCA5KQk7Froh1meP2PpoIl4fDNI6TnO7TqCQ6u25Gm/smvt+o2oXr8J5ixYoiibu3AZXFp6wK19Fxw4fEyp/tETJzHsl7F5HeZXa/V9N1T7tqnKbfq8hQCAuYuXoVGr9mjZ4QfsP6Lc5yMnTmHoqHH5EfZX27JtJ1q374K6DZuju/cA3An8b1udu2ApXJq7w829Ew4cOqr0uKPHT2LYCN+8DjdXZGbMf4hPwMFVm7Ggty9m/DAUy4dOxtXDp5XaOeK3HXN+HIGFfcbi9qlLSsvunruKzdOX5Ul/ckJRGPcc80V3zOcmjfwOIDeULl0amzdvxvz586GrqwsASEhIwMaNG1GmTBmluo8fP0aDBg1gYmKC6dOno3LlytDW1sbt27excuVKlCxZEu3atcuz2M/tPIyrh0/DY0gPWJSxxouHYdi7ZD109HVQp00TAMDPa2YqPebhtUDsXbYB5etVBwBcO3IG4Y+eoNeMX/DoWiB2zl+L4X6zIZPJ8PbVG1w/eha954zJsz5lV+Dde/h79z44Odoryk6dOYdDR49j2YLZePL0GSZPn4Nv6taGqYkx3sfEYMmKtfhj4Zx8jDp7/lq1DHK5XHH/4eMQ9P95FJo3dsGpM+dw8OgJLJ8/C0+ePsekGXNQv85HfV65BisWFL4+Hz56AvMWLsO40cNRqWJ5bNy8HQOH/YJdW//E7cAgHDp8DMsWzcGTJ88x+bdZ+KZebZiamKT2eflq/LFkXn53IUdkZswf8f8bobeD0f6nXjApbo7HN+7iwMrNMDA1hnOdqrh/+RbuBFyG54ShiAx/jb1L/4RD9QrQMzJAQmw8/vlrD7wmDcvnnmZOURn3HPNFd8znJknM1H2qRo0aKF26NHbs2KEo27FjB8qUKYPq1asr1R04cCA0NDRw5coVdO7cGeXLl4e9vT08PDywf/9+uLu752nsz4Ifw7lOVTjVqgyT4uaoUL8G7KuVx/MHYYo6BqbGSrfgy7dgW6ksTK0sAABvnr1E2dpVULyMNWq1ckHcuxjEvYsBABxcsQlNu38HbT3dPO1XVsXFxWPs5OkYP2YEjAwNFeUhoU9Qs3o1VCzvjFYtmkJfXx8vXoQDABYuXYlO37mjhJVlfoWdbWamJihmbqa4nT53AaVLWqNW9ap4HPYEtapXRcVyzmjVvAn09fXwPDy1zwuWrUSn79oVyj5v2LQNHTzawMO9FRzsbTFuzHDo6Ohg194DCAkNQ80a1VCxfDm0atkU+nr6ePHiJQBg4eIV6NTRo1D2OT2ZGfPP7j1CFdd6sK1UFibFzVGjRUNY2pbEi4ehAFLHvE1FJ1g72qBSw9rQ1tVB1Ks3AIDj63egplsjGFuY5Uf3sqQojXuO+cI95gvqHkJJJnUA4O3tDT8/P8X9tWvXolevXkp1IiIicOTIEQwaNAj6+vrptiOTyXI1zk+VcrZHyK17iHjxCgDwMuQZngY9gmP1iunWj4l6h4dXb6Na0/qKMkvbUnga9AhJiR/w6MZdGJgaQ8/IALdPXYK6libK1auWF135KjPmLUTD+nVRr3ZNpfKyjg4IuheMd+/e4+69+0hMTETpUiVx/eZtBAU/QNdOHfIp4pyTlJSEA0eOwaONG2QyGZwdHXD33v2P+vwBZUr+v8/3H6Lb99/ld8hZlpSUhKB7wahb57/1q6amhrq1a+LW7bso6/TReg4K/m8937iFoOD76Nq58K/nNJkZ86XKOeD+5Vt4FxEFIQRCbwcj8sVr2FetACB1zIc/eoL4mFiEPwpD0ockmJawwJOghwh//BR1WjfOl75lVVEd9xzzhX/Mp+0hjI+PV5R9bg9h9erVceTIEUyfPh3Xr1/H+fPnMWrUKOzbtw/Hjh37tPlMk+TuVwDw8vKCr68vwsJSv+2ePXsWmzdvxsmTJxV1Hj58CCEEnJ2dlR5brFgxJCQkAEjNsmfNmpVncTfo0BKJ8QlYNmQy1NRkkMsFGndrh8ouddKtf+ufC9DS1VHsegWAak3r43XYM/wxbAp0DQ3QcWRvJMTE4dTmveg+9Wf8s3E3As9chalVMbgP+hFG5iZ51LvMOXT0BO4FP8CGNctVltWvVxutWzaDl88AaGtrY8r40dDV1cH0OQsw+dfR2LZzDzZv3wkTY2OMHz0cDvZ2+dCDr3Pi9Fm8j4lBu9YtAQD169ZG6xbN4NlnILS1tTF1XGqff5u3EFPGjsK2XXuxaftOmJgYY/wvw+Fob5u/HciEt1HRSEmRw8xMefbI3MwUoWFPUL9eHbR2aw6vXv1S1/NE39T1PHs+Jo8fg207dmPz1v/32XdEoVzPaTIz5t16d8b+5X9hYR9fqKmrQSZTQ5sBnrCp6AQAcKheAZUb1cGaUbOgoaUJjyE9oKWtjQMrNsFjyI+4evg0Lh34B3qGBmgzwBPFy1jnV3czVJTHPcd84R/zNWrUwKNHj7Bjxw54enoC+G8PoZ2dcqwf7yH8eEIpbS/hx8fTZpVkkzoLCwu0adMG/v7+EEKgTZs2KFasWKYee+nSJcjlcnh6eiIxMTHDeomJiSrLtbW1vyruwHNXcef0ZXz3cy9YlLbGq5BnOLJ2GwzNjFG18Tcq9W+cOIfKDetAQ0tTUaauoY5Wfbsq1duzeD1qt2mMl4+fIvjiTfT9fRzO7TqCw2u2oNOofl8Vc056+eo15ixYiuULZ0NbWyvdOv1790T/3j0V91esWYe6tWtAQ0Mdq/03YOufaxBw9jzGT52JjX4r8ijynLNr/0E0qFsHxT/aXgf49MAAnx6K+3+sXY+6tVL7vGrdBmxbtxqnz13A+GkzsWntH/kRdo7r36cX+vf5b3Z9xWp/1K1dExoaGli99k9s3eiHgDPnMX7SDGxcvzIfI/06mRnzl/efxLP7IejiOwDGFmZ4cvchDq3aDEMzY9hXLQ8AcPmhLVx+aKto99SWfbCvUg5q6uoI2H4Q/eb/igdXbmP3In/0mVuwTioo6uOeYz5Vfo/5jD7TM/u5nraHMC2pS9tD+PFkUtoewunTp+fKHkLJ7n4FUl9gf39/rFu3Dt7e3irLHR0dIZPJEBwcrFRub28PR0dHxUkWGZkxYwaMjY2VbjNmzPiqmI+v24n6HVqg0re1YWlTElVc66KuexOc3XFYpe6Tuw8Q8fwVqjVr8Nk2Q28H49+nL1C7lSvCAh/AsWYlaOloo0L9mgi78+Cr4s1pQffuI/LtW3Tr1Q+1GjZDrYbNcPX6TWzathO1GjZDSkqKUv2Q0CfYf/gYBvbxxpVrN1GjWhWYmZqgRVNXBAU/QGxs3p69/LVevHyFi1eu4Tv31hnWCQl7gv1HjmFQ7164cv0malRN7XPLJi4Iuv8AsXl8xnZ2mJoYQ11dDZGRkUrlEZFvYW6meuxXSGgY9h88ioH9vHHl2g3UqF41dT03c0VQ8P1Ct54/9qUxn5T4ASc27kaLnt+jbO0qsLQthdqtXVGhQU1c2J3+bpo3z17izulLcO3qjrDA+yhTwRH6xoao0KAmXj5+isT4hLzs4hcV5XHPMV9wxvzXfqZ7eXnhzJkzCAsLQ1hYGM6ePQsvLy+lOp/bQ2hgYAADAwOMHj06232Q7EwdALi5ueHDhw+QyWRo2bKlynJzc3M0b94cS5YswZAhQzLMmjPi6+uL4cOHK5Vpa2tj28Oz2Y45KfGDSpYuU1ODkKtOx14/fg4lHMrAyq5Uhu0lf0jCwVWb0f4nb6ipq0HI5ZD/f2pXnpKidPZVQVCnVg1s+3ONUtnE32bDzqY0enp1hbq6uqJcCIFps3/HiKEDoKenC7k8BcnJyQCA5OTUD4GC1r8v2b3/EMxMTdDwm3rpLhdCYNqc+Rg5uD/09HSRIpcr+pz0/z6npBT8PmtqaqJ8OWdcvHwNjV0aAkhdV5cuX0WXTsrHCwkhMG3m7xjx0yDo6elB/lGf0/7K5cof+oXJl8a8PCUF8uQUyNSU66ipqaW7m0YIgf1/bETznt9DS1cHcrmAPG08/D85EgVsXBTlcc8xX3DGfEaf6ZmVF3sIv0TSSZ26ujqCgoIU/6dn2bJlaNCgAWrVqoVJkyahSpUqUFNTw+XLl3Hv3j3UrFkz3ccBWZuWzSyn2pVxZvshGBczg0UZa7x8/BQX9x5H1Sb1leolxsUj6Nw1NO/Z8bPtnd52AI41KqGEfWkAqQdcH1+3A1WbfIPLB06idDmHHI3/a+nr68HRQfn4A11dHRgbG6mU79yzH6YmJnD5NvW1qValElasWY9bd+7i7IVLsLezgaGhQZ7F/rXkcjn2HDgEd7cW0NBIf3vdsfcATE2M/+tz5UpYsTa1z2cuXoK9rQ2MCkmfvbp2woQpM1ChvDMqVUi9vEF8QgI82rZSqrdz9/7UPjf8aD2v8set24E4e/4S7O1sYfjRmZKFzZfGvLaeLmwqOuHYuh3Q0NJK3f0a+AC3Tl1Md/xfP3YWekYGKFu7CgCgdDkHnN6yD8+CH+Ph9UAUK10COvp6edrHLymq455jvmCN+Zz4TPf29sbgwYMBAEuXLlVZ/rk9hAC+uIfwSySd1AGAkZHRZ5c7ODjg+vXrmD59Onx9ffHs2TNoa2ujQoUKGDlyJAYOHJhHkaZy690FJzfuwcGVmxH77j0MTY1Ro8W3aNSpjVK9wDNXIIRAxW9rZ9jW67DnCDp3FX3m/XeRygrfVEfYnftY9+s8mFtb4rufVXdLFwYRkZFYve4v+K9YrCirVKE8vLp2wtCRY2FmaoIp47M/hZ0fLly5hvBXr9G+jVu6yyMiI7F6/V9Y98ciRVnlCuXQ/YfvMWTUWJiZmmLKuMLT55bNm+BtVBSWr/RDREQknMs6YumC2TA3/29XTEREJFb7/wn/Vf+9OVaqWB5e3Tpj6HBfmJmZYMqEwn1B0syM+Q7DfXBiw27sWrAW8TFxMLYwQ+Nu7VCzZSOltmKi3uHM9oPoNeMXRVlJJ1vUa9cMm39bBj1jQ3gM7YHCSmrjnmNeemM+t/cQfolMfM1pFpSuDYEn8juEPONVsQniIp7ndxh5Ss+8JOL/fZbfYeQpXYtSiIsKz+8w8oyeSYks1eeYlzaO+aIhK+O+Z8+eiIqKwq5du5T+B4B3794B+G9SqX379jAxMYG/vz8A4NGjR2jQoAFMTU1V9hCOHDkSnp6emDcvexdalvxMHREREVFeyc89hJypywX81i5t/NYufZypyxjHfNFQ1MY8kPVxXxBJ+pImREREREUFkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEiATQoj8DoKIiIiIvo5GfgcgRXFR4fkdQp7RMymBv25szu8w8pRntR+K1DoGUtdzUeqznkmJLNUvaq8Nx7z0FbUxD2R93BdE3P1KREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJgGSSOnd3d7i5uaW7LCAgADKZDLdu3YJMJsONGzcAAKGhoZDJZIqboaEhKlasiEGDBuHBgwd5GP3nXb1+E8NG+KJ5m46oXtcV/5wKUFq+fsNmNHFrjyZu7bH+ry1Ky27fuYtuP/ZFcnJyXoacJYnxiTjsfxALB/2O6V5TsXb8ajx/+Fyx/ENCIg6u3Y/5A+ZhutdULBu+BFeOXlZq4/D6Q5jtPRMLBs7D7YBbSsvung/Epll/5Ulfskvq6zgjW7btROv2XVC3YXN09x6AO4FBimVzFyyFS3N3uLl3woFDR5Ued/T4SQwb4ZvX4eYZqW8PHPPSX8cZ4ZjPXZJJ6nx8fHD06FE8e/ZMZZmfnx9q1aoFIyOjdB977NgxhIeH4+bNm5g+fTqCgoJQtWpVHD9+PLfDzpT4+ASUdXKA7y8/qSy7/+ARlq/0w8xpEzBj6ngsW7EGDx4+BgAkJyfjt1m/Y9yY4dDQ0MjjqDNv74rdeHz7EdoP6oD+cwfCvooDNkxbh3eR7wAAR9YfxsMbD/Hd4A4Y+Ptg1G1dDwfXHkDwlXsAgOCrwbhz5ja8xnVHM88W2LtiN+LexQIAEuIScGLLcbT2aZNv/csMqa/j9Bw+egLzFi5DP5+e2LhuFco6OmDgsF8QGfkWpwLO4dDhY1i2aA6GDe6PKdPn4G1UFADgfUwMlixfjTHpvFZSIfXtgWNe+us4PVIY8wV9AkkySV3btm1hYWEBf39/pfKYmBhs27YNPj4+GT7W3NwcVlZWsLe3h4eHB44dO4a6devCx8cHKSkpuRz5l31bvy4G9e+NJq4NVZaFhj2Bk6M96tSqgbq1a8LJ0QGhYU8AAOs2bEGNalVRsUK5vA4505I+JCHoYhCaeraATQVbmFmZw7VTY5hZmeHKkdRv5k+Dn6KqS1XYVrSDSXFT1GxWC1Y2lopv9m+e/QvbCrawdiiJSg0qQ1tPG29fRwEAjm04glrNa8G4mEk+9TBzpLyOM7Jh0zZ08GgDD/dWcLC3xbgxw6Gjo4Ndew8gJDQMNWtUQ8Xy5dCqZVPo6+njxYuXAICFi1egU0cPlLCyzOce5B4pbw8c86mkvI4zIoUxX9AnkCST1GloaODHH3+Ev78/hBCK8m3btiElJQVdu3bNdFtqamoYNmwYwsLCcPXq1dwIN8c4Otgj7OkzhL98hRfhLxH25Ckc7O3w9Nlz7Nl3EIP6Z5zMFgTyFDmEXA4NTeVvnBpamnganPomVtq5NO5fCca7yHcQQiDkTggiwiPgUMUBAGBpY4UXj18gPiYeLx6/QNKHZJhZmeHJvTC8DAlHnVb18rxfOamwr+P0JCUlIeheMOrWqakoU1NTQ93aNXHr9l2UdXJA0L1gvHv3HneDgpGYmIjSpUri+o1bCAq+j66dO+Rj9PmrsG8PHPNfVtjXcXqkMuYL+gRS4Zq7/QJvb2/MmTMHp06dgqurK4DUzLljx44wNjbG27dvM91WuXKp34JCQ0NRp06ddOskJiYiMTFRqUxbWzt7wWeTvZ0NBg/ojQFDRgIAhgzsA3s7G/QbPBw/DemHcxcuYcVqf2hoaOCX4UNQs3rVPI3vS7R1tVGqbGkE7DgFi5LFoG9igDtnb+PZ/acwszIDALj1ao19K/dgwYB5UFNXg0wmQ9u+7WBTwRYA4FjNEZUbVsHqsSuhqaWB9gO/g5aOJg6s3od2A7/DlSOXcfnQRega6qFt33YoXrp4PvY46wr7Ok7P26hopKTIYWZmplRubmaK0LAnqF+vDlq7NYdXr37Q1tbGlIm+0NXVwfTZ8zF5/Bhs27Ebm7fuhImJMcb7joCDvV0+9STvFfbtgWP+ywr7Ok5PYRjzGX2mf/y5/vEE0rhx4yCTyQAoTyBlNtdIm0D67rvvcPXq1QxzjayQVFJXrlw51K9fH2vXroWrqysePnyIgIAATJkyJcttpc32pa2w9MyYMQOTJ09WKps4cSJG/dQvy8/3NTp18ECnDh6K+3v2H4K+nh6qVKqI9p27Y4PfCrx+/S/G/DoF+3dugpaWVp7G9yXtB3XAnj92Yf6AeZCpqaGEXQlUalAZ4Y9fAAAuHbqI5w+eocuobjApZoywoDAcXLsfhqaGsP//N3fXTo3h2qmxos1T2/6BXWV7qKurIWDHafSfOxAPrt7H7qU70Gdm/3zp59co7Os4O/r36YX+fXop7q9Y7Y+6tWtCQ0MDq9f+ia0b/RBw5jzGT5qBjetX5mOkea+wbw8c819W2NdxduT3mM/oM33SpElKZXk9gZQVktn9msbHxwd///033r9/Dz8/Pzg4OMDFxSXL7QQFpZ6RY2eX8bcBX19fREdHK918ffP37Jy3UVFYuXodRo8YituBQbApUxo2ZUqhdq3qSE5ORtgT1eMA8puZlRl6TvLGmHXj8NOy4eg9vS9SUlJgYmmKpA9JOLHpOJr/6Abnms6wtLFCHbe6qPhNJZzfdy7d9t48/xe3z9xC4y5NEHo3FDblbaBvpI8K31REeEg4EuMT031cYVEY1/GnTE2Moa6uhsjISKXyiMi3MP/kmzwAhISGYf/BoxjYzxtXrt1AjepVYWZqghbNXBEUfB+xsXF5FXqBUxi3B475rCmM6/hThWHMZ/Yz/eMJJACKCaTP7XrNSGYmkLLiq5K6bdu2wcPDA2XLlkWpUqWQkJCAqVOnYsqUKXjz5k2OBJhVnTt3hpqaGjZu3Ij169fD29s7yy+WXC7HokWLYGdnh+rVq2dYT1tbG0ZGRkq3vN79+ql585fCs2snWFoWh1wuVzrlPSUlBXJ5/p/4kREtHS0YmhoiPiYej24+gnOtcpAnp0CekqKyDmVqMqVjJ9MIIbBv1V40/9ENWjraEHK54lgF+f//Crk89zuTiwrzOk6jqamJ8uWccfHyNUWZXC7HpctXUaVyBaW6QghMm/k7Rvw0CHp6ekp9TvtbGPqcWwrz9sAxnzmFeR2nKQxjPiuf6Xk5gZQV2dr9KoSAp6cntmzZorgvk8mgo6ODAwcO4NKlSyhWrBgGDhyYI0FmhYGBAbp06QJfX1+8e/cOPXv2/OJjIiIi8PLlS8TFxeHOnTtYsGABLl26hP3790NdXT33g/6CuLg4PH323zWcnr94ieD7D2BkZKR0NtCFi1cQ9vQZpkxM/WZRsbwzQsOe4My5i3j16jXU1dRgU6ZMnsf/JQ9vPAQgYG5dDJEvI3FswxEUsy6Gaq7Voa6hDpsKtji24Qg0tTRgbGGCsLuhuHX6Jlr82FKlresnrkLfSB/ONZ0BAKWdy+DUtpN4dv8pHt54AItSFtDR183jHn6Z1Ndxery6dsKEKTNQobwzKlUoj42btyM+IQEebVsp1du5ez9MTYzh0rA+AKBalUpYscoft24H4uz5S7C3s4WhoWF+dCHXSH174JiX/jpOj5TGfOfOnTFs2DDFBNKAAQNybQIpK7KV1C1evBibN29Od1nr1q1x8eJF7Nq1K1+SOiA1g16zZg1at24Na2vrL9Zv1qwZAEBPTw82NjZo3LgxVq5cCUdHx9wONVPuBgWjz8CfFffnLVgKAHBv0xJTJqQO9ISERMycuxCzfpsANbXUCVhLy+IYNWIoJk2dCU0tLUyZ4AsdnfydSUxPYnwCTmw6hncR76BroIvydSug8Q9Noa6RmlB3HPY9jm88hp2L/0Z8TDyMLUzQ+IemqNm8tlI7MVExCNgZAO8p/02Bl3QshXpt62PTrL+gb6QPj0Hf5WnfMkvq6zg9LZs3wduoKCxf6YeIiEg4l3XE0gWzYW7+366YiIhIrPb/E/6rlirKKlUsD69unTF0uC/MzEwUr4+USH174JiX/jpOj5TGfEGdQJKJ9Oazv6BatWq4ffs26tWrB29vb/Tp0wcymQwpKSnYsmULunbtChsbG4SEhORIkIVNXFR4foeQZ/RMSuCvG+kn+FLlWe2HIrWOgdT1XJT6rGdSIkv1i9prwzEvfUVtzANZH/fnz59H/fr10bp1a+zfv19RHhoaCjs7O1y/fh3VqlVT3Fc8z0cTSD///HOOTiBla6bu/v37AIBx48bB2NhYaZmFhQUA4OXLl18ZGhEREVHB9M0336R7nKetra1S+af3c1O2TpTQ1NQEkHqxvU+l/eSFrm7BO4aBiIiISKqyldRVrlwZADBp0iTFb5sBwOnTp/Hbb79BJpOhWrVqOREfEREREWVCtpI6Hx8fCCEQHByMoUOHKs74aNy4seL30LJzvRYiIiIiyp5sJXW9evVC9+7dIYRQ2k+c9v+PP/4IT0/PnImQiIiIiL4o2z8Ttm7dOri7u+Ovv/5SnDhRtmxZeHp64vvvv8+xAImIiIjoy77qt1+///57JnBEREREBcBXJXV37tzB7t278fjxYwCAvb09PDw8UKlSpRwJjoiIiIgyJ1tJXUpKCoYMGYIVK1aoLJswYQL69OmDZcuWKa6ATURERES5K1tZ1+TJk/HHH38oTpT49LZq1SpMmjQph0MlIiIiooxkK6lbs2YNAEBdXR2enp5YuHAhFi1aBE9PT6irq0MIoahDRERERLkvW7tfo6KiIJPJMHnyZIwdO1ZRPnjwYJQvXx6//voroqOjcyxIIiIiIvq8bM3U1apVCwDS/dWItLK6detmOygiIiIiyppsJXXz5s2Dnp4elixZgvfv3yvKY2JisGTJEhgbG2P+/Pk5FiQRERERfV62dr+OGjUKpqamOHz4MKytrVGuXDkAQHBwMGJjY2FjY4OffvpJ6TEymQzHjx//6oCJiIiISFW2krqTJ08qfu81NjYW165dA5D6M2EymQxhYWEICwtT1E8rJyIiIqLcke2LD6f3m6+f/k9EREREeSNbSV1ISEhOx0FEREREXyFbSV2pUqWgrq6e07EQERERUTZl6+xXKysrDBw4EKdPn87peIiIiIgoG7KV1EVERGDFihVo3LgxSpcujV9++UVxsgQRERER5b1sJXXm5uaK33l9/vw5fv/9d9SuXRvOzs6YPHkygoODczpOIiIiIvqMbCV1r169wunTpzFy5Eg4OzsrErwHDx5gypQpqFChAmrUqIF58+bhxYsXOR0zEREREX0iW0mdmpoavv32W8yePRtBQUG4f/8+5syZgxo1aigSvJs3b2LUqFGwtbXFgAEDkJCQkNOxExEREdH/Zfs6dWlSUlIQHByMK1eu4N69e4qLDKddry45ORkrV66Empoali5d+rVPR0RERETpyNZMHQCcO3cOgwYNQokSJdCuXTts3boVcXFxEELA0tISo0ePxqlTp9ClSxcIIbB9+/acjJuIiIiIPpKtmTp7e3vFz4ClzchpaGigdevW8PHxQevWrRXXsXN2dsaWLVvw5s2bHAqZiIiIiD6VraQuNDRU8X/ZsmXh7e2NHj16wNLSUqWukZERGjVqxN9+JSIiIspF2Urq9PT00KlTJ/j4+ODbb7/9bF0dHR2cPHkyO09DRERERJmUraRu8+bNMDIyQtWqVXM6HiIiIiLKhmydKNGuXTs0adIEt2/fVll29uxZqKurQ0Pjq0+sJSIiIqJMynbmlXaCxKdSUlIyXEZEREREuSPTSd2TJ0+UTpAAgOvXryM5OVlxXy6XY8OGDakNc6aOiIiIKM9kOvPy8/PDlClTFPeFEBg6dGi6dWUyGWxsbL4+OiIiIiLKlCxNp326W/Vzu1kHDBiQvYiIiIiIKMsyndRVq1YNPXr0AACsW7cOMpkMbm5uKF68uKKOmpoaTE1N0bhxY7Rp0ybnoyUiIiKidGU6qfPw8ICHhweA1KQOAMaNG4f69evnTmRERERElGnZOpshJCQEAGBtbZ2jwRARERFR9mQrqeNJEEREREQFS7YuPkxEREREBYtM8ErBRERERIUerxCcC+JfP8vvEPKMbvFSiHv7Ir/DyFN6ptbYcGd/foeRp7wqtUFcVHh+h5Fn9ExKZKk+x7y0ccwXDVkd9wURd78SERERSUCOJXVJSUk51RQRERERZVG2k7rk5GTMmTMHVatWhba2NnR1dZGQkAAfHx94e3vj6dOnORknEREREX1Gto6pS0hIgJubGwICAgCk/lyYTCaDjo4OwsLC8M8//6BChQoYOXJkjgZLREREROnL1kzd7Nmzcfr0aQghVH7/tXnz5hBCYO/evTkSIBERERF9WbaSuo0bN0Imk6Ft27YqyZujoyOA/351goiIiIhyX7Z2v4aGhgIAhgwZAj09PaVlJiYmAIDXr19/VWBERERElHnZmqlLS+RevFC9VtGtW7cAAEZGRl8RFhERERFlRbaSupo1a0IIgXHjxuHQoUOK8vXr12Pq1KmQyWSoXbt2jgVJRERERJ+XraRu8ODBAIDw8HBMnz4dMpkMANCrVy9ERUUp1SEiIiKi3JetpM7DwwO//vqr4uzXj28AMH78eLRq1SpHAyUiIiKijGX7t1+nTJmCdu3a4a+//sL9+/cBAGXLlkW3bt2465WIiIgoj2U7qQOAWrVqoVatWjkVCxERERFlU7aSuidPnmSqXpkyZbLTPBERERFlUbaSOltbW8XJERmRyWRITk7OVlBERERElDXZ3v366c+DEREREVH+yVZS16hRI5WZujdv3uDevXuQy+UoVaoUHBwcciRAIiIiIvqybCV1J0+eTLc8NDQUrVu3xvPnz7FgwYKvCIuIiIiIsiJb16nLiK2tLQYOHIj3799j5MiROdk0EREREX1GjiZ1KSkpOH36NADg3LlzOdk0EREREX1Gtna/2tvbq5SlpKQgIiIC8fHxAABDQ8Ovi4yIiIiIMi1bSV1oaGi6lzT5+IxYHx+f7EdFRERERFmSo5c0MTY2hqOjI/r27YvevXt/VWBERERElHnZSurkcnlOx0FEREREXyHLSV1cXBwGDx4MmUwGDw8PtGvXLjfiIiIiIqIsyHJSp6enh82bNyMxMRGdO3fOjZiIiIiIKIuydUmTqlWrAgAiIyNzNBgiIiIiyp5sJXWzZ8+GtrY2Jk2ahIcPH+Z0TERERESURdk6UWLixIkwMzPDgwcPUL58eTg5OcHS0lLpMicymQzHjx/PsUCJiIiIKGPZ/u1XmUwGmUyGlJQUBAcHIzg4WLFcCJHudeyIiIiIKHdkOqlL+/mvatWqAVC+Tl1616wjIiIioryT6aTO1dUVampqOH36NEJCQnIzJiIiIiLKoiztfk2bkbOxscmVYIiIiIgoe7J19isRERERFSxZPlHi+vXrSE5OzlTdRo0aZTkgIiIiIsq6LCd1Q4cOzVQ9mUyW6eSPiIiIiL5OlpM6nulKREREVPBkOamzsrKCtrZ2bsRCRERERNmU5aRu+/btqF+/fm7EQkRERETZVKjPfj1//jzU1dXRpk0bpfK0X7yIiopSeYytrS0WLFiguJ/2yxgymQz6+vpwcnJCz549cfXq1VyOPnNSUlKwdLUfWnf2RN2mrdC2ixdW+v+ptBt83aataOzeEY3dO2L95q1Kj78dGISuPv2RnJyS16Fn25p1f8GzV380aNIaTVp9h59H/YrQsCdKdeYuWAqXFu3g1q4zDhw6qrTs6PGTGDZibF6GnGWJ8Qk4vHYnFvWbihldR8Fv7CK8ePhfH4Mu3MJfU/7A3B6/YmrH4XgZ8lyljSN+uzGnxzgs7DsFt08rb693z93A5umrc70fX2vLtp1o3b4L6jZsju7eA3AnMEixbO6CpXBp7g43904ZrGPfvA43T3DMc8xzzBetMZ+TCnVSt2bNGgwZMgSnT5/Gixcvst2On58fwsPDERgYiKVLlyImJgZ169bF+vXrczDabMb212Zs27UHY34agh0b/DCsfx/4b9yCTX/vBADcf/gIy9f4Y9akXzFz4jgsXeWHB48eAwCSk1Mwbd4CjBv5EzQ01POzG1ly7fpNdOnYHutXL8XyRXOQnJyMAcNGIT4+HgBwKuAcDh05jmUL52DY4H6YMmMu3kZFAwDex8RgyR9rMOaXYfnZhS/at2wrHt+8D4+h3dDv919gX7UsNkz+A+8iogAASQkfULqcHZp2b5vu4+9fDsSdM9fgOb4/mnZvi33LtyDuXQwAICE2Hv9sPIhWfTrmVXey5fDRE5i3cBn6+fTExnWrUNbRAQOH/YLIyLep6/jwMSxbNAfDBvfHlOlz8Pb/X9Lex8RgyfLVGPPLT/kaf27hmOeYTw/HvHTHfE7KdFJXpkwZlClTBjo6OrkZT6bFxMRgy5YtGDBgANq0aQN/f/9st2ViYgIrKyvY2tqiRYsW2L59Ozw9PTF48GC8ffs254LOhpt3AuH6bX00ql8PJUtYoXljF3xTpxbu3L0HAAh58hRODvaoU7M66taqAScHe4Q8eQoAWLdpC2pUrYxK5cvlZxeybOmC2WjX1g0O9nZwdnLE5PFj8PLlK9y9dx8AEBIahpo1qqFieWe0atEU+nr6ePEiHACwcMkKdOrQDiWsLPOzC5+VlPgBQRduodmP7rCp6ACzEhZw6eIGU6tiuHr4HACgimstNOrcEnZVyqbbxpvnr2BT0QHWjqVRqWENaOvqIOp1JADg+J97UbNlfRhbmOZZn7Jjw6Zt6ODRBh7ureBgb4txY4ZDR0cHu/Ye+Ggdl0Orlmnr+CUAYOHiFejU0aNAr+OvwTHPMZ8ejvmCN+YL4t7CTCd1oaGhCAkJQY0aNbL1RDlt69atKFeuHJydneHl5YW1a9fm6Jm5P//8M96/f4+jR49+uXIuqlqpIi5evY6w/79pBz98hOu3bqNBvToAACd7O4Q9fYbwV6/w4uUrhD19Bkc7Wzx9/gK7DxzC4D7e+Rl+joiJiQUAGBsZAQDKOjkg6F4w3r17j7v3gpGYmIjSpUri+o3bCAp+gK6dO+RnuF8kl8sh5HJoaCof0qqppYmn9zL3E3yWNtYIf/QU8TFxCH/0FEkfkmBqVQxPgh4j/PFz1GndMDdCzzFJSUkIuheMunVqKsrU1NRQt3ZN3Lp9V3kdB328jm8hKPh+gV/HX4NjnmM+PRzzBW8dF8S9hVk+UaKgWLNmDby8vAAAbm5uiI6OxqlTp+Dq6poj7Zcrl/pNNzQ0NEfayy5vr66IjYtDe69eUFdTQ4pcjsF9vNGmRTMAgL2tDYb09UH/n0cBAIb26w17Wxv0++kX/DSgL85dvIw//NZDQ0MDo4YOQs1qVfKzO1kml8sxd8ESVKtSCY4OdgCA+vXqoHXL5vDy7g9tbW1MmTAGuro6mD5nPiaPH41tO/Zg87adMDE2wnjfEXCwt8vnXijT1tVBKWdbBGw/imKlLKFvbIg7Z67h2f1QmFoVy1QbDtXLoXKjmlgzej40tDThMaQrtLS1cGDldngM7oqrh8/i0sEz0DPUR5v+nVG8jFUu9ypr3kZFIyVFDjMzM6VyczNThIY9SV3Hbs3h1atf6jqe6Ju6jmfPx+TxY7Btx25s3roTJibGBXIdfw2OeY759HDMF6wxn7a38MqVK3j58iX8/f0xdmz2jutM21sIQLHHsEePHhg8eDDc3d1hapr5GdhCmdQFBwfj0qVL2Lkz9RgTDQ0NdOnSBWvWrMmxpC5t1k8mk2VYJzExEYmJiUplOX25lyMnTuLA0eOYMWEsHOxsEfzgEeYsXgqLYuZo16olAKBTe3d0au+ueMyeg4ehp6eLqpUqwMOzJ/5auQyv//0XYyZNw/6tG6ClpZWjMeamGXMW4uGjEPitXKxU3r9PT/Tv01Nxf8XqdahbuyY01DWw2u9PbP1rLQLOnsf4yTOwcd3KPI76yzyGdsPepZuxoM9kyNTUUMK+JCp+Wx3hj55lug2XLm5w6eKmuH9q62HYVykLNXV1BPx9DP1+/wUPrt7F7sUb0WfO8NzoRq7q36cX+vfppbi/YrV/6jrW0MDqtX9i60Y/BJw5j/GTZmDj+oK3jrOLY55jPiMc87k75jP6TE/vc/3TvYU//fQTfH19P5szZMXPP/+M9evX4+jRo+jcuXOmH1coT5RYs2YNkpOTYW1tDQ0NDWhoaGD58uX4+++/ER0dDaP/T9lHR0erPDYqKgrGxsZffI6goNQzcuzsMv42MGPGDBgbGyvdZsyYkc1epW/+8pXo5fkD3Jo1gZODPdq6NYdX5++xdsOmdOu/jYrGCr8/MeanIbh99x5sSpeCTelSqF2jOpKTkxH2NPNvIPlt5tyFCDh7HquWzYdlcYsM64WEPsH+Q0cxsK83rly7gRrVq8DM1AQtmroiKPgBYmPj8jDqzDGzKoYeUwdj9F8zMGzlBPjM+hnyZDlMLc2z1d6bZ69w59RVuP7QCmGBD1GmvD30jQ1QoX5VvHz8DInxCTncg69jamIMdXU1REZGKpVHRL6F+Sff5IHUY6r2HzyKgf3S1nHV1HXczBVBwfcL5DrOLo55jvnM4JjP+XWclc/0jPYW5pTs7i0sdEldcnIy1q9fj3nz5uHGjRuK282bN2FtbY1NmzbByckJampqKgcaPn78GNHR0ShbNv0DUT+2YMECGBkZoVmzZhnW8fX1RXR0tNLN1zdnT7lOSEiAmkx5NampqUEul6dbf+7iZfDq3BGWxS0gl8uVLmuQnJKClAweV5AIITBz7kKcOHUGK5b8jpLWJT5bd9qseRgxbCD09HSV+pz2N6PXqiDQ0tGGoakR4mPi8OjGPTjXrpTlNoQQ2L9iG5r39ICWrjbkcgF5yv/7npLadyEvWL8Eo6mpifLlnHHx8jVFmVwux6XLV1GlcgWlukIITJv5O0b8NAh6enr/X8epP0GY9lcuLzyX7/gSjnmO+S/hmM+dMZ/Zz/S0vYVdu3YFoLy3MKdkZm9hegrd7td9+/bh7du38PHxUZlx69ixI9asWYP+/fujd+/eGDFiBDQ0NFC5cmU8ffoUo0ePRr169VQunhwVFYWXL18iMTER9+/fx4oVK7Br1y6sX78eJiYmGcaS0bRsfI70NFWj+t9g9Z9/wcqy+P93xTzEhi3b4dHGTaXu+ctXEPb0GaaOGw0AqFjOGaFhT3DmwkW8fP0v1NXVYFumdA5GlztmzFmAg0eOY/7sadDX18ObiNRvdgb6+tDRUX69d+7eD1MTE7g0TF2n1apUworV63Drzl2cPX8R9nY2MDQ0yPM+fMmj6/cgIGBuXRxvX77BsfV7UaxkcVRtknowfPz7WES/icL7yNTZ5ogXrwEABiaGMDA1Umrr+rEL0DMyQNnaFQEApcvZ4vTWw3h2PxQPr91DsVKW0NHXzcPeZY5X106YMGUGKpR3RqUK5bFx83bEJyTAo20rpXqp69hYeR2v8set24E4e/4S7O1sYWhomB9dyBUc8xzzAMd8foz5jD7TP/Xx3sI0Qghoa2tjyZIlSnsLP80hcnJvYXoKXVK3Zs0aNGvWLN0XpWPHjpg9ezZu3bqFhQsXYubMmRg9ejTCwsJgZWWF5s2b47ffflPJfHv1St2Hr6Ojg5IlS+Lbb7/FpUuXCsSZvmN+HoKlq/0w4/eFiHwbBYti5ujo0Rb9enZXqpeQmIiZ8xdj1uTxUFNL/ZZvWdwCo38ajIkz5kBLUxNTxo6GTiH4ibdtO/YAAPoM/FmpfPKvo9Gu7X8fbBERkVjtvwH+q5YoyipVLA+vbp0wdLgvzExNMGXCmLwJOosS4hLwz1/78S4iCroGeihXrwoad2sN9f9fW+z+5UDsWbpZUX/H738CABp1bqF0TE1M1Huc+fsYek0fqigr6WSDeu4u2PzbaugZG8BjSNc86lXWtGzeBG+jorB8pR8iIiLhXNYRSxfMhrn5f7tiUtfxn/BftVRRlrqOO6euYzMTTJkgrQuScsz/h2OeYx4oWGP+472FLVq0UFrWvn17bNq0CZ6enoq9hTY2NorlOb23MD0ykZPXASEAQPzrwnMMy9fSLV4KcW+zfyp3YaRnao0Nd/bndxh5yqtSG8RFhed3GHlGzyTj3X/p4ZiXNo75oiEz437Xrl3o0qULXr9+rTK5NHr0aJw4cQKXL19Gv379cOTIESxcuFBpbyEAnDt3TjG5JJPJ4OfnBzc3t3T3Fnbr1i1LfSh0M3VERERE+aGg7y1kUkdERESUCXv37s1wWZ06dZR+BGHSpEmYNGnSZ9vL6Z2lhe7sVyIiIiJSxaSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQTIhBAiv4MgIiIioq+jkd8BSFFcVHh+h5Bn9ExKFKn+AkW3zxuvb8jvMPJMt+peWapflLaHorr9F8U+F6UxD2R93BdE3P1KREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJQKFP6nr27AmZTKZyc3Nzw4sXL2BqaopFixYpPebixYvQ1NTEkSNHAAD+/v7ptqGjo5MfXUrXlm070bp9F9Rt2BzdvQfgTmCQYtncBUvh0twdbu6dcODQUaXHHT1+EsNG+OZ1uDmiqPVZSv0NCwrDptmbMW/AfEz+YSruXb6ntFwIgX+2nsS8/vPxW/cZWD9tAyLCI5TqxMfEY8finZjRaxZmes/G7j/24kPCB8XyqNdR8Ju0DtN7zITfpHWIeh2l9PiNszbj7sUgFFZS2h4yq6j1WUr95ZgvGAp9UgcAbm5uCA8PV7pt2rQJ1tbWWLx4MXx9ffHgwQMAQHx8PHr06IHevXujRYsWijaMjIxU2ggLC8uvLik5fPQE5i1chn4+PbFx3SqUdXTAwGG/IDLyLU4FnMOhw8ewbNEcDBvcH1Omz8HbqCgAwPuYGCxZvhpjfvkpX+PPjqLWZ6n190NCEixtLNG6V6t0l5/dcw4XD11Cm96t0XuaN7S0NbFhxkYkf0hW1NmxeCdeP/sX3cd6oduoH/Dk3hPsXblPsfzIhqMwMjNEv5l9YGhigCMb/vvgu3MuEDI1oELd8rnXyVwkte0hM4pan6XW36I25gvqhJIkkjptbW1YWVkp3UxNTQEAXl5eaNmyJXr27Am5XA5fX18kJSVhzpw5Sm3IZDKVNiwtLfOjOyo2bNqGDh5t4OHeCg72thg3Zjh0dHSwa+8BhISGoWaNaqhYvhxatWwKfT19vHjxEgCwcPEKdOrogRJWBaMfWVHU+iy1/jpVd0STLo1Rvk45lWVCCFw8eAmNvmuIcrWcYWljifaDPPD+7Xvcu5L67f7f5//i4c1HaNe3LUo5lUSZcmXQqmdL3DkfiPeR7/9f5w2qNqoC8xLmqOpSFW9evAEAJMQm4J+tJzP8cCkMpLY9ZEZR67PU+lsUx3xBnFCSRFL3JX/88QcePHgAT09PLFmyBH5+fjAwMMjvsDIlKSkJQfeCUbdOTUWZmpoa6tauiVu376KskwOC7gXj3bv3uBsUjMTERJQuVRLXb9xCUPB9dO3cIR+jz56i1uei1t+o11GIiYqBfWU7RZmOng5KOZbE0/vPAQDP7j+Hjr4OrB2sFXXsK9tDJpPh2cPUOlY2lnh8OwRCLvDo1mNYlkn9kDvy1zHUblELxsWM87BXOaeobQ9A0etzUeuvVMd8QZxQ0viqHhUQ+/btU0nSxo4di7FjxwIAihcvjqlTp6J///4YMGAAGjVqpNJGdHS0ShsNGzbEwYMHM3zexMREJCYmKpVpa2tntxvpehsVjZQUOczMzJTKzc1MERr2BPXr1UFrt+bw6tUP2tramDLRF7q6Opg+ez4mjx+DbTt2Y/PWnTAxMcZ43xFwsLfL4JkKjqLW56LW35ioGACAvrG+Urm+sT5i/78sJioG+kZ6SsvV1NWga6CreHxzr2bYt+oAFgxZBMsylmjbpzXCgsLwKvQlmndrim0LtuPF43A4VLFHq55uUNdQz4Pefb2itj0ARa/PRa2/hWXMZ/SZnt3P9T/++AOVKlWCp6cntm3bhhMnTuT6hJIkkrrGjRtj+fLlSmUfD5aUlBT4+/tDT08PFy5cQHJyMjQ0lLtuaGiIa9euKZXp6up+9nlnzJiByZMnK5VNnDgRo37ql51uZFv/Pr3Qv08vxf0Vq/1Rt3ZNaGhoYPXaP7F1ox8CzpzH+EkzsHH9yjyNLbcUtT4Xtf5mhpGZEbqN/kFxPzkpGRumb0T7gR44vSMAWjraGPz7QPw1YyOuHLuKum518jHanFUUt4ei1uei1t/MyO0xn9Fn+qRJk9Ktn18TSp8jid2v+vr6cHR0VLp9nNTNnTsXjx8/xpUrV/Ds2TNMnz5dpQ01NTWVNkqWLPnZ5/X19UV0dLTSzdc3Z89IMjUxhrq6GiIjI5XKIyLfwvyTb3kAEBIahv0Hj2JgP29cuXYDNapXhZmpCVo0c0VQ8H3ExsblaHy5oaj1uaj118Ak9Q0sNjpWqTw2Ohb6/19mYGKA2HfK/ZCnyBEfE694/KcCdp2BQxV7WNuXQOjdMFSoWw7qGuooV6ccwu4WjJOeMqOobQ9A0etzUetvYRnzWf1Mb9y4MW7cuKF069+/v2J5ehNKnzI0NFRpY/Xq1VmOPY0kkrrPCQwMxMSJE7F8+XKUL18ey5cvx7Rp03Dr1q2vbltbWxtGRkZKt5ze/aqpqYny5Zxx8fJ/s4hyuRyXLl9FlcoVlOoKITBt5u8Y8dMg6OnpQS6XKzaitL9yeUqOxpcbilqfi1p/TYqbwMDEAI/vhCjKEuMS8ezhc5Qum/pFqlTZkkiITcCLx+GKOiF3QiCEQClH1S9b/z7/F3fOBqJxZ1cAgJDLkZIiB5D6wSCXy3OxRzmrqG0PQNHrc1Hrb2EZ81n9TM+vCaXPkcTu18TERLx8+VKpTENDAyYmJujRowc6dOiADh1SDyzt2LEjOnbsiJ49e+LSpUuK3bBCCJU2gNTpUzW1/M19vbp2woQpM1ChvDMqVSiPjZu3Iz4hAR5tlc/02bl7P0xNjOHSsD4AoFqVSlixyh+3bgfi7PlLsLezhaGhYX50IcuKWp+l1t8PCR8Q+fK/WYi3r6PwMvQldA10YVzMGHVb1UHAzjMwtzKDSXET/LP1JAxNDVGuVuqZcxYlLeBY1QF7V+5D296tkZIixwG/Q6j0TUUYmin3TwiBfSv3o2X35tDS0QIAlHYujWvHr8O8hBlunr6FSg0q5l3nc4DUtofMKGp9llp/OeaVpU0obdy4UTGh1LVrV7Rv3x5VqlTJteeVRFJ36NAhlChRQqnM2dkZ3bp1w/PnzxXXhEmzdOlSVKxYEdOnT8eECRMAAO/evVNpAwDCw8NhZWWVe8FnQsvmTfA2KgrLV/ohIiISzmUdsXTBbJib//eNICIiEqv9/4T/qqWKskoVy8OrW2cMHe4LMzMTTJlQsC5W+TlFrc9S6++LRy+wbuqfivtH/ky9nlTVRlXQfqAHGrSrj6TEJOxdtR8JcQko41wGXmO6QUPrv7ekDkO+w4G1B7F+2gbIZDKUr1sOrXq6qTzX1ePXoG+sj7I1yyrKXL93wd+Ld2L1r2vhWNUBdVrUzsXe5jypbQ+ZUdT6LLX+FsUxXxAnlGRCCJGNvtBnxEWFf7mSROiZlChS/QWKbp83Xt+Q32HkmW7VvbJUvyhtD0V1+y+KfS5KYx7I2rjv2bMn1q1bp1KeNqG0fPlyBAYGKu2OjYyMRMWKFTFgwABMmDAB/v7+6NWrl0obQPYnlJjU5YKiNPiL6ptdUexzUXqDZ1KXsaK6/RfFPhelMQ9kfdwXRJI/UYKIiIioKGBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BERERFJAJM6IiIiIglgUkdEREQkAUzqiIiIiCSASR0RERGRBDCpIyIiIpIAJnVEREREEsCkjoiIiEgKBBV6CQkJYuLEiSIhISG/Q8kzRa3PRa2/QhTNPmdWUXxt2GfpK2r9zQ0yIYTI78SSvs67d+9gbGyM6OhoGBkZ5Xc4eaKo9bmo9Rcomn3OrKL42rDP0u9zUetvbuDuVyIiIiIJYFJHREREJAFM6oiIiIgkgEmdBGhra2PixInQ1tbO71DyTFHrc1HrL1A0+5xZRfG1YZ+lr6j1NzfwRAkiIiIiCeBMHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BVQPXv2RPv27RX/y2Qy9O/fX6XeoEGDIJPJ0LNnT6Xyly9fYtiwYXB0dISOjg4sLS3RoEEDLF++HHFxcXnQg6xzd3eHm5tbussCAgIgk8lw69YtyGQy3LhxAwAQGhoKmUymuBkaGqJixYoYNGgQHjx4kIfRZ9358+ehrq6ONm3aKJWfPHkSMpkMUVFRKo+xtbXFggULFPc/7ru+vj6cnJzQs2dPXL16NZejz560bfnTm5ubG168eAFTU1MsWrRI6TEXL16EpqYmjhw5AgDw9/dPtw0dHZ386FKO4ZhXxjGfimNeumM+NzCpKyRKly6NzZs3Iz4+XlGWkJCAjRs3okyZMkp1Hz9+jOrVq+PIkSOYPn06rl+/jvPnz2PUqFHYt28fjh07ltfhZ4qPjw+OHj2KZ8+eqSzz8/NDrVq1MvzpmGPHjiE8PBw3b97E9OnTERQUhKpVq+L48eO5HXa2rVmzBkOGDMHp06fx4sWLbLfj5+eH8PBwBAYGYunSpYiJiUHdunWxfv36HIw257i5uSE8PFzptmnTJlhbW2Px4sXw9fVVfDjHx8ejR48e6N27N1q0aKFow8jISKWNsLCw/OpSruCY55jPCMe8NMd8TtDI7wAoc2rUqIFHjx5hx44d8PT0BADs2LEDZcqUgZ2dnVLdgQMHQkNDA1euXIG+vr6i3N7eHh4eHiioV7Fp27YtLCws4O/vj19//VVRHhMTg23btmHOnDkZPtbc3BxWVlYAUvvp7u6Opk2bwsfHB48ePYK6unqux58VMTEx2LJlC65cuYKXL1/C398fY8eOzVZbJiYmir7b2tqiRYsW6NGjBwYPHgx3d3eYmprmZOhfTVtbWxHvp7y8vLBjxw707NkTAQEB8PX1RVJSksq6l8lkGbYhFRzzHPMZ4ZinjHCmrhDx9vaGn5+f4v7atWvRq1cvpToRERE4cuQIBg0apPTm/jGZTJarcWaXhoYGfvzxR/j7+yt9CG3btg0pKSno2rVrpttSU1PDsGHDEBYWViB3S2zduhXlypWDs7MzvLy8sHbt2hz94P3555/x/v17HD16NMfazCt//PEHHjx4AE9PTyxZsgR+fn4wMDDI77DyBcc8x3xmccwTwKSuUPHy8sKZM2cQFhaGsLAwnD17Fl5eXkp1Hj58CCEEnJ2dlcqLFSsGAwMDGBgYYPTo0XkZdpZ4e3vj0aNHOHXqlKLMz88PHTt2hLGxcZbaKleuHIDUY3AKmjVr1ijWnZubG6Kjo5X6/LUKct/37dun2BbTbtOnT1csL168OKZOnYrNmzejb9++aNSokUob0dHRKm20atUqL7uRJzjmOeYzqyD3nWM+73D3ayFiYWGBNm3aKL7VtmnTBsWKFcvUYy9dugS5XA5PT08kJibmcqTZV65cOdSvXx9r166Fq6srHj58iICAAEyZMiXLbaV9Cy5osxTBwcG4dOkSdu7cCSB1tqJLly5Ys2YNXF1dc+Q5CmrfAaBx48ZYvny5UpmZmZni/5SUFPj7+0NPTw8XLlxAcnIyNDSU36oMDQ1x7do1pTJdXd3cCzqfcMxnTUHd7jnmOebzCpO6Qsbb2xuDBw8GACxdulRluaOjI2QyGYKDg5XK7e3tARSOQeDj44MhQ4Zg6dKl8PPzg4ODA1xcXLLcTlBQEACoHH+U39asWYPk5GRYW1sryoQQ0NbWxpIlSxQHhkdHR8PExETpsVFRUZmavSiofQcAfX19ODo6Zrh87ty5ePz4Ma5cuQIXFxdMnz4dEyZMUKqjpqb22TakhGM+8wrqds8xzzGfV7j7tZBxc3PDhw8fkJSUhJYtW6osNzc3R/PmzbFkyRLExsbmQ4Rfr3PnzlBTU8PGjRuxfv16eHt7Z/nbp1wux6JFi2BnZ4fq1avnUqRZl5ycjPXr12PevHm4ceOG4nbz5k1YW1tj06ZNcHJygpqamspxQY8fP0Z0dDTKli37xedZsGABjIyM0KxZs9zqSq4IDAzExIkTsXz5cpQvXx7Lly/HtGnTcOvWrfwOLd9wzGcOxzzHPHGmrtBRV1dXfCPL6OyuZcuWoUGDBqhVqxYmTZqEKlWqQE1NDZcvX8a9e/dQs2bNvAw5ywwMDNClSxf4+vri3bt3KtfjSk9ERARevnyJuLg43LlzBwsWLMClS5ewf//+AnUW3L59+/D27Vv4+PiofPvu2LEj1qxZg/79+6N3794YMWIENDQ0ULlyZTx9+hSjR49GvXr1UL9+faXHRUVF4eXLl0hMTMT9+/exYsUK7Nq1C+vXr1f51l8QJCYm4uXLl0plGhoaMDExQY8ePdChQwd06NABQOpr0rFjR/Ts2ROXLl1S7JIRQqi0AaQem6OmJq3vqhzz6eOY55gHpDnmv4qgAqlHjx7Cw8ND5f/0eHh4iB49eiiVvXjxQgwePFjY2dkJTU1NYWBgIOrUqSPmzJkjYmNjcy/wHHLu3DkBQLRu3VqpPCQkRAAQ169fV7qfdtPT0xPly5cXAwcOFA8ePMiHyD+vbdu2Kn1Kc/HiRQFA3Lx5U8THx4uJEyeKcuXKCV1dXWFnZyf69u0r/v33X6XHfNx3HR0d4eDgIHr06CGuXr2aF93Jsh49eijFnHZzdnYWkydPFlZWViIiIkLpMREREcLKykpMnjxZCCGEn59fum0AEOHh4fnRrRzBMc8xzzGfqqiM+dwgE6KAXsCIiIiIiDKNc5ZEREREEsCkjoiIiEgCmNQRERERSQCTOiIiIiIJYFJHREREJAFM6oiIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQBTOqIiIiIJIBJHREREZEEMKkjIiIikgAmdUREREQSwKSOiIiISAKY1BER5RJ/f3/IZDLIZDJMmjQpv8MhIonTyO8AiKjwSkhIgL+/P/7++2/cvHkTUVFRMDU1hbW1NerWrYvvvvsOLVq0gEwmy+9QC4Vdu3bhxo0bAICePXvC1tY2X+MhosKFSR0RZcv9+/fh4eGBe/fuKZW/fv0ar1+/xo0bN7BixQq8f/8eBgYG+RRl4bJr1y6sW7cOAODq6sqkjoiyhEkdEWVZVFQUWrZsidDQUACAubk5hg4dirp160JNTQ3379/H/v37cfjw4fwNlIioCOExdUSUZXPnzlVK6C5fvowJEyagZcuWaN68OQYNGoQDBw7g9u3b0NbWVjwu7fiyT2egXF1dFcvS2k2ze/duNGvWDKamptDW1oazszMmT56M+Ph4pXqhoaHo1q0brK2toampCRMTE1SoUAG9evXCrVu3FPVu3rwJDw8PFC9eHJqamjA3N0e1atXQv39/PHnyRKnNgIAAtGvXDhYWFtDS0oKdnR2GDx+Ot2/fqrwmJ06cQO3ataGjowMHBwcsXbo0069naGgoZDKZYpYOABo3bqx4Tfbu3Qt9fX3FayeEUNRLSUmBhYUFZDIZzM3NkZSUpGhPJpPB1dUVly9fhouLC/T09GBtbY3x48cjOTlZKQYhBPz8/NCgQQMYGRlBV1cXVatWxcKFCyGXyzPdFyLKR4KIKIvs7e0FAAFAzJgxI9OPS3uMjY2NUrmLi4tiWUhIiKJ8/PjxivJPbw0bNhSJiYlCCCGSkpJE2bJlM6y7atUqIYQQb968ERYWFhnWO3r0qOK5V61aJdTU1NKt5+zsLCIjIxV1z549K7S0tFTqValSRfH/xIkTM3xdQkJCMowJgPjnn39Ejx49FPcDAgIUjz19+rSivG/fvirtlSpVSujr66u02a9fP6UYfvzxxwyfv0uXLplex0SUfzhTR0RZEhMTg8ePHyvuN2nSRPF/eHg4zpw5o3T7dPYrsy5fvoypU6cCAEqUKIE1a9bg0KFDaNOmDYDUWbT58+cDAO7du4f79+8DAJo1a4ZDhw5h3759WLx4MVq1aqWYLTx//jz+/fdfAEDXrl1x9OhR7Nq1C3PnzoWLiwvU1dUBAM+fP8fgwYMhl8thaGiIxYsX4/Dhw+jVqxcAIDg4GGPHjlXEOmLECHz48EHx/Hv37sXUqVMRGBiYqb6WKFECAQEBaNWqlaJs0aJFCAgIQEBAAKpXrw4fHx/Fsr/++kvx/549exT/d+3aVaXtZ8+eoUGDBoqY0vq4YsUKxQzm9u3bsX79egCAs7MzNm3ahL1796JevXoAgC1btmDLli2Z6gsR5aP8ziqJqHB59uyZ0ixOcHCwYtnixYtVZnk+nqFKK8vMTN2wYcMUZWPHjhUBAQEiICBA7N27V1FeqVIlIYQQ9+7dU5R1795dPHr0SKSkpKjEfujQIUW9UaNGiSdPngi5XK5Sb/78+Yp6vXr1Ujz36dOnhZ6engAgjI2NRUpKinj16pWirra2toiIiFC04+npmamZujQfz8b9888/KsvTZiPNzc3Fhw8fhBBCODs7CwDC2tpa0eePZ+r09PREVFRUujFNmTJFCCGEh4eHomzRokWK/q5atUpR3rZt2y/GT0T5izN1RJQlxsbGSvefPXuWK8+TNvMGANOnT0fDhg3RsGFDuLu7K8rTzrx1cnJCw4YNAQB//vknHBwcYGBggG+++QZz5sxBYmIiAKBhw4ZwcnICAMyePRtlypSBsbExXF1dsWrVKsWxYx8/t5+fn+K5GzVqhLi4OABAdHQ0Xrx4oTRr6eDgADMzM8X9OnXq5Ohr4u3tDQCIiIjAoUOH8ODBAwQHBwMAunTpAjU11bf0cuXKKa2zj2NKi/3j/g4dOlTR3z59+ijKg4KCcrQvRJTzmNQRUZYYGBjA3t5ecf/cuXOK/wcPHgwhBEaPHv3ZNlJSUpTuv3nzJluxJCcnIzExEWpqajhw4ADmzZsHNzc3lClTBvHx8bhw4QJGjRqFYcOGAQD09PRw9uxZTJkyBU2aNIGVlRXev3+PU6dOoW/fvpg9e3aWnj82Nvazy3P6+nw9evSAhkbqRQs2bNiA3bt3K5Z169YtU21kN6Yv9ZWI8h+TOiLKsi5duij+nzdvHl68eJGpx6XNGEVERCApKQlA6pmfn17rDgDKli2r+N/Pzw9CCJVbbGwstLW1IYSAgYEBhg8fjoMHDyIsLAyvX7+GnZ0dAGDHjh0AUs/wtLCwwPjx43H8+HGEh4fj8ePHiuvopdX7+LknTpyY4XM7OzsrngNInfn6+MzYixcvZup1SfPxTFt6Z5xaWVmhdevWAIC9e/di06ZNAABHR0fUqlUr3TaDg4Px7t27dGNKS84/7u8///yTbn8fPXqUpb4QUd7jdeqIKMtGjhyJv/76C0+ePEFUVBRq166N4cOHo3r16khISMCVK1fSfZyjoyOuXr2K+Ph4dOvWDY0aNcKyZctUZu6A1JmnhQsXAgB+/vlnREZGokqVKoiKisKjR49w5MgR2NjYYO3atXj+/DmaNWuGzp07o0KFCrC0tERISIjipIi03a/nzp3D0KFD0bFjRzg5OaFYsWK4deuWYpdqWr3vv/8eY8aMQWJiImbOnAmZTIZvvvkGcXFxCAkJwT///IP4+HgcPXoUlpaWqFu3Li5evIiEhAT88MMPGDp0KG7evInNmzdn6XU1NTVV/L9hwwaoq6tDXV0d3377raLcx8cHe/bsQXx8PK5duwYg/RMk0sTGxqJLly4YPHiwSkweHh4AAE9PT8WsX/fu3TFu3Dg4OTnh33//xYMHD7B//360atUKEydOzFJ/iCiP5fVBfEQkDYGBgUqXNsnoNm3aNMVjVqxYobLcwMBAlCpVKsuXNAEgevToIYQQ4unTp5+tl3b5joCAgM/W+/jyLJ+7pAkA4eLioqh7+vRpoampqVLHyckpSydKfHwSyMe3jyUlJQkrKyul5Xfv3lWq8/GJEjY2NsLIyEilzd69eys95nOXNMls/ESUv5jUEVG2xcTEiPnz54uGDRsKMzMzoa6uLoyMjETVqlVFv379xMGDB5XOQk1JSRG+vr6iePHiQldXVzRp0kRcv349w+vUCSHEvn37hJubmzA3NxeampqiZMmS4ttvvxUzZ84UoaGhQgghYmNjxcSJE4WLi4soUaKE0NTUFLq6uqJKlSpi2rRpijNFX716JUaPHi3q1asnLC0thYaGhjAwMBC1a9cWS5cuVTkT9syZM6JDhw6KupaWlqJOnTpi/PjxIjAwUKnu0aNHRY0aNYSWlpawsbERs2bNEmvXrs1yUjR37lzh4OAgNDQ00k3qhBBi9OjRimVVq1ZVWf5xUufi4iKuX78uXF1dha6urrCyshJjx44VSUlJKo9bv369cHFxEcbGxkJLS0uUKVNGNG3aVCxatEi8evUqU/ETUf6RCfHRpcmJiKjAO336NFxcXAAAs2bNwqhRo5SWh4aGKo71c3FxwcmTJ/M6RCLKBzymjoiokIiPj8e7d++wfPlyAIC6unqmz3olIuljUkdEVEi0atUKp06dUtz39vZGqVKl8jEiIipImNQRERUyxYoVQ8eOHfH777/ndyhEVIDwmDoiIiIiCeDFh4mIiIgkgEkdERERkQQwqSMiIiKSACZ1RERERBLApI6IiIhIApjUEREREUkAkzoiIiIiCWBSR0RERCQB/wNmtqXu3RHl3gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compareHeatmaps(xgbG8, xgbRecG8, xG8_valid, yG8_valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## LightGBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S] Validation accuracy: 0.96\n",
      "[S] Train accuracy: 0.9999074074074074\n",
      "[S] [Recommended] Validation accuracy: 0.9625\n",
      "[S] [Recommended] Train accuracy: 0.9997222222222222\n",
      "\n",
      "\n",
      "[G4] Validation accuracy: 0.9483333333333334\n",
      "[G4] Train accuracy: 0.991574074074074\n",
      "[G4] [Recommended] Validation accuracy: 0.9575\n",
      "[G4] [Recommended] Train accuracy: 0.9944444444444445\n",
      "\n",
      "\n",
      "[G6] Validation accuracy: 0.9233333333333333\n",
      "[G6] Train accuracy: 0.9880555555555556\n",
      "[G6] [Recommended] Validation accuracy: 0.9308333333333333\n",
      "[G6] [Recommended] Train accuracy: 0.9930555555555556\n",
      "\n",
      "\n",
      "[G8] Validation accuracy: 0.9391666666666667\n",
      "[G8] Train accuracy: 0.9887037037037038\n",
      "[G8] [Recommended] Validation accuracy: 0.9441666666666667\n",
      "[G8] [Recommended] Train accuracy: 0.9929629629629629\n",
      "\n",
      "\n",
      "[G4S] Validation accuracy: 0.9633333333333334\n",
      "[G4S] Train accuracy: 0.9993518518518518\n",
      "[G4S] [Recommended] Validation accuracy: 0.9658333333333333\n",
      "[G4S] [Recommended] Train accuracy: 0.9994444444444445\n",
      "\n",
      "\n",
      "[G6S] Validation accuracy: 0.95\n",
      "[G6S] Train accuracy: 0.9988888888888889\n",
      "[G6S] [Recommended] Validation accuracy: 0.9525\n",
      "[G6S] [Recommended] Train accuracy: 0.9993518518518518\n",
      "\n",
      "\n",
      "[G8S] Validation accuracy: 0.9575\n",
      "[G8S] Train accuracy: 0.9989814814814815\n",
      "[G8S] [Recommended] Validation accuracy: 0.96\n",
      "[G8S] [Recommended] Train accuracy: 0.9993518518518518\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "showAccuracy('lgbm')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compareHeatmaps(lgbmG8, lgbmRecG8, xG8_valid, yG8_valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## CatBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[S] Validation accuracy: 0.955\n",
      "[S] Train accuracy: 0.999537037037037\n",
      "[S] [Recommended] Validation accuracy: 0.9533333333333334\n",
      "[S] [Recommended] Train accuracy: 0.9993518518518518\n",
      "\n",
      "\n",
      "[G4] Validation accuracy: 0.9466666666666667\n",
      "[G4] Train accuracy: 0.986574074074074\n",
      "[G4] [Recommended] Validation accuracy: 0.9425\n",
      "[G4] [Recommended] Train accuracy: 0.9783333333333334\n",
      "\n",
      "\n",
      "[G6] Validation accuracy: 0.9175\n",
      "[G6] Train accuracy: 0.9840740740740741\n",
      "[G6] [Recommended] Validation accuracy: 0.9141666666666667\n",
      "[G6] [Recommended] Train accuracy: 0.9641666666666666\n",
      "\n",
      "\n",
      "[G8] Validation accuracy: 0.9316666666666666\n",
      "[G8] Train accuracy: 0.9831481481481481\n",
      "[G8] [Recommended] Validation accuracy: 0.9333333333333333\n",
      "[G8] [Recommended] Train accuracy: 0.9717592592592592\n",
      "\n",
      "\n",
      "[G4S] Validation accuracy: 0.965\n",
      "[G4S] Train accuracy: 0.9987037037037036\n",
      "[G4S] [Recommended] Validation accuracy: 0.9466666666666667\n",
      "[G4S] [Recommended] Train accuracy: 0.9544444444444444\n",
      "\n",
      "\n",
      "[G6S] Validation accuracy: 0.9491666666666667\n",
      "[G6S] Train accuracy: 0.9983333333333333\n",
      "[G6S] [Recommended] Validation accuracy: 0.945\n",
      "[G6S] [Recommended] Train accuracy: 0.9860185185185185\n",
      "\n",
      "\n",
      "[G8S] Validation accuracy: 0.9433333333333334\n",
      "[G8S] Train accuracy: 0.9985185185185185\n",
      "[G8S] [Recommended] Validation accuracy: 0.9416666666666667\n",
      "[G8S] [Recommended] Train accuracy: 0.9823148148148149\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "showAccuracy('cbc')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "compareHeatmaps(cbcG8, cbcRecG8, xG8_valid, yG8_valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ✍ Testing\n",
    "_[> Back to contents](#Contents:)_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import dill\n",
    "dill.load_session('models.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "tS   = pd.DataFrame(columns = Split)\n",
    "tG4  = pd.DataFrame(columns = NGrams)\n",
    "tG6  = pd.DataFrame(columns = NGrams)\n",
    "tG8  = pd.DataFrame(columns = NGrams)\n",
    "tG4S = pd.DataFrame(columns = Split)\n",
    "tG6S = pd.DataFrame(columns = Split)\n",
    "tG8S = pd.DataFrame(columns = Split)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "File: 100%|▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬| 16/16 [01:08<00:00, 4.28s/it]\n"
     ]
    }
   ],
   "source": [
    "path  = \"data_test/\"\n",
    "files = [ file for file in listdir(path) if isfile(path + file) ]\n",
    "        \n",
    "for file in tqdm(files, desc='File'):\n",
    "    \n",
    "    hexfile = hexlify(open(path + file, 'rb').read()).decode()            # hex representation\n",
    "    Bytes = [ hexfile[i:i+2] for i in range(0, len(hexfile)-1, 2) ]       # \n",
    "    \n",
    "    partSize = len(Bytes) // 3\n",
    "    part1 = Bytes[            :   partSize]\n",
    "    part2 = Bytes[   partSize : 2*partSize]\n",
    "    part3 = Bytes[ 2*partSize :           ]\n",
    "\n",
    "    tS   =   tS.append( pd.DataFrame([ calc(Bytes, part1, part2, part3, frmt, mode=\"S\"  )], columns = tS.columns   ))\n",
    "    tG4  =  tG4.append( pd.DataFrame([ calc(Bytes, part1, part2, part3, frmt, mode=\"G4\" )], columns = tG4.columns  ))\n",
    "    tG6  =  tG6.append( pd.DataFrame([ calc(Bytes, part1, part2, part3, frmt, mode=\"G6\" )], columns = tG6.columns  ))\n",
    "    tG8  =  tG8.append( pd.DataFrame([ calc(Bytes, part1, part2, part3, frmt, mode=\"G8\" )], columns = tG8.columns  ))\n",
    "    tG4S = tG4S.append( pd.DataFrame([ calc(Bytes, part1, part2, part3, frmt, mode=\"G4S\")], columns = tG4S.columns ))\n",
    "    tG6S = tG6S.append( pd.DataFrame([ calc(Bytes, part1, part2, part3, frmt, mode=\"G6S\")], columns = tG6S.columns )) \n",
    "    tG8S = tG8S.append( pd.DataFrame([ calc(Bytes, part1, part2, part3, frmt, mode=\"G8S\")], columns = tG8S.columns ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['conversation.mp3',\n",
       " 'documents.bmp',\n",
       " 'door.exe',\n",
       " 'google_dino.png',\n",
       " 'map.bmp',\n",
       " 'opening.wav',\n",
       " 'puppy-attacks-toy.jpg',\n",
       " 'puppy-meme.png',\n",
       " 'python-3.7.4.exe',\n",
       " 'recording.avi',\n",
       " 'secret.bmp',\n",
       " 'steps.avi',\n",
       " 'task.wav',\n",
       " 'they.mp3',\n",
       " 'tryharder.exe',\n",
       " 'wise-puppy-in-glasses.jpg']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.875"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(leType.transform(['Audio','Image','Executable','Image','Image',\n",
    "                                'Audio', 'Image','Image','Executable',\n",
    "                                'Video','Image','Video','Audio',\n",
    "                                'Audio','Executable','Image'\n",
    "                                ]), knnRecG8.predict(tG8.values))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Audio', 'Image', 'Executable', 'Image', 'Image', 'Audio', 'Image',\n",
       "       'Image', 'Executable', 'Executable', 'Image', 'Video', 'Audio',\n",
       "       'Executable', 'Executable', 'Image'], dtype='<U10')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "leType.inverse_transform(knnRecG8.predict(tG8.values))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 📚 Resources\n",
    "_[> Back to contents](#Contents:)_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Datasets:\n",
    "\n",
    ">`[1]` Chao Dong, Chen Change Loy, Xiaoou Tang. Accelerating the Super-Resolution Convolutional Neural Network, in Proceedings of European Conference on Computer Vision (ECCV), 2016 [arXiv:1608.00367](https://arxiv.org/abs/1608.00367)\n",
    "<br>\n",
    "\n",
    ">`[2]` Zhang Zhifei, Song Yang, and Qi Hairong. \"Age Progression/Regression by Conditional Adversarial Autoencoder\". IEEE Conference on Computer Vision and Pattern Recognition (CVPR). [arXiv:1702.08423](https://arxiv.org/abs/1702.08423), 2017\n",
    "<br>\n",
    "\n",
    ">`[3]` H. Kuehne, H. Jhuang, E. Garrote, T. Poggio, and T. Serre. HMDB: A Large Video Database for Human Motion Recognition. ICCV, 2011. [PDF](http://serre-lab.clps.brown.edu/wp-content/uploads/2012/08/Kuehne_etal_iccv11.pdf) [Bibtex](http://serre-lab.clps.brown.edu/wp-content/uploads/2013/10/Kuehne_etal_iccv11.bib)\n",
    "<br>\n",
    "\n",
    ">`[4]` Eduardo Fonseca, Manoj Plakal, Daniel P. W. Ellis, Frederic Font, Xavier Favory, and Xavier Serra, “Learning Sound Event Classifiers from Web Audio with Noisy Labels”, arXiv preprint [arXiv:1901.01189](https://arxiv.org/abs/1901.01189), 2019\n",
    "<br>\n",
    "\n",
    ">`[5]` Royi Ronen, Marian Radu, Corina Feuerstein, Elad Yom-Tov, Mansour Ahmadi. \"Microsoft Malware Classification Challenge\". [arXiv:1802.10135](https://arxiv.org/abs/1802.10135), 2018"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Images:\n",
    "\n",
    ">`[6]` [An Audio Encryption Using Transposition Method](https://www.researchgate.net/publication/281444376_An_Audio_Encryption_Using_Transposition_Method) - Scientific Figure on ResearchGate.\n",
    "<br>\n",
    "\n",
    ">`[7]` Mp3filestructure, original by [Kim Meyrick](https://en.wikipedia.org/wiki/User:Kim_Meyrick). Traced by [Stannered](https://commons.wikimedia.org/wiki/User:Stannered) (23 Apr 2007), fixed by [Mik](https://commons.wikimedia.org/wiki/User:Mik) (20 Dec 2013). \n",
    "<br>\n",
    "\n",
    ">`[8]` Portable Executable 32 bit Structure, original by ByteBiter. Fixed by [Benhut1](https://commons.wikimedia.org/wiki/User:Benhut1) (1 Sep 2016), fixed by [Finnusertop](https://commons.wikimedia.org/wiki/User:Finnusertop) (29 Sep 2019). "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}