Algorithm Selection.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "PGvaUSn4HNfS"
   },
   "source": [
    "#### Preparation\n",
    "This project can use Weights & Biases for tracking metrics. Enabling this requires an account and api key, which can be set below. If this key is ommitted, metrics will not be pushed to Weights & Biases, but all analyses will work just fine.   \n",
    "\n",
    "Weights & Biases can be installed using: `pip install wandb`.  \n",
    "\n",
    "This notebook requires the execution of:\n",
    "- `Data Exporation and Preprocessing` "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 255
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 20096,
     "status": "ok",
     "timestamp": 1584530640018,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14GiEvCunvq33iGZ3RU-P9BMF7L60e8LoQ3Y1AfZJ4Q=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "Gu0xcYjlHVRi",
    "outputId": "1a66507c-2e43-4b35-b95d-1f59b2c2f538"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "# Optional: set your W&B api key below. Looks like: 7ol65g3j2rulhg3af4o7m8wma8d9isyu3myqkdb8 (example)\n",
    "os.environ['WANDB_API_KEY'] = ''\n",
    "os.environ['WANDB_PROJECT'] = 'particle_process_discriminator'\n",
    "\n",
    "if os.environ['WANDB_API_KEY'] is not '':\n",
    "    # login to W&B account\n",
    "    res = os.system('wandb login {}'.format(os.environ['WANDB_API_KEY']))\n",
    "    print(\"Logged in to Weights & Biases\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TCBcBszIuCwV"
   },
   "source": [
    "#### Initialize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-02-01T16:10:35.460693Z",
     "start_time": "2020-02-01T16:10:33.168375Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 12166,
     "status": "ok",
     "timestamp": 1584530717405,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14GiEvCunvq33iGZ3RU-P9BMF7L60e8LoQ3Y1AfZJ4Q=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "i65FQNgLuCwb",
    "outputId": "d3094531-50c7-40a5-e15a-309c33b088ef"
   },
   "outputs": [],
   "source": [
    "# (basic) dependencies\n",
    "import os\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "from tqdm import tqdm as tqdm\n",
    "\n",
    "# sklearn dependencies\n",
    "from sklearn.metrics import auc\n",
    "from sklearn.model_selection import train_test_split, KFold\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score\n",
    "from sklearn.model_selection import cross_val_score\n",
    "\n",
    "# tensorflow dependencies\n",
    "from tensorflow.keras import datasets, layers, models, regularizers, backend\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.layers import LeakyReLU\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Dropout, LSTM, Embedding\n",
    "\n",
    "if os.environ['WANDB_API_KEY'] is not '':\n",
    "    import wandb\n",
    "    from wandb.keras import WandbCallback\n",
    "\n",
    "# import classifiers\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.ensemble import BaggingClassifier\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "\n",
    "MODEL_DIR = 'models'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6oxFdseXGzyr"
   },
   "outputs": [],
   "source": [
    "# own dependencies\n",
    "from helper import metrics, learn, select_features, load_data, cross_validation, ROC\n",
    "from helper import make, save, make_args, build_architecture, train_net, train_args"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Eug86HrfG5CJ"
   },
   "outputs": [],
   "source": [
    "# further settings\n",
    "FIG_DIR = 'figures/'\n",
    "plt.rcParams[\"figure.figsize\"] = (10, 4.5)\n",
    "plt.rcParams[\"savefig.format\"] = \"pdf\"\n",
    "plt.rcParams[\"savefig.dpi\"] = 400\n",
    "plt.rcParams[\"savefig.transparent\"] = True\n",
    "plt.rcParams.update({'font.size': 14})\n",
    "plt.rcParams[\"savefig.bbox\"] = 'tight'\n",
    "\n",
    "random_state = 576"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "weLgX8qUrKTT"
   },
   "outputs": [],
   "source": [
    "def plot_accuracy_gain(acc, labels, baseline, fn=None, label_rotation=0):\n",
    "    f, axes = plt.subplots(1, 1, figsize=(10, 5))\n",
    "    ax = axes\n",
    "\n",
    "    # calculate means and std. between trails\n",
    "    means, stds = np.mean(acc, axis=-1), np.std(acc, axis=-1)\n",
    "    ax.bar(range(len(labels)), height=means - means[baseline], yerr=stds, capsize=4)\n",
    "    ax.axhline(0, color='k')\n",
    "    ax.set_xticklabels(labels, rotation=label_rotation)\n",
    "    ax.set_xticks(range(len(labels)))\n",
    "    \n",
    "    # Set xticks to percentage\n",
    "    vals = ax.get_yticks()\n",
    "    labels = ['{:,.1%}'.format(x) for x in vals]\n",
    "    ax.set_yticklabels(labels)\n",
    "    ax.set_ylabel(\"Validation accuracy gain\")\n",
    "\n",
    "    f.tight_layout()\n",
    "    if fn is not None:\n",
    "        f.savefig('{}/{}'.format(FIG_DIR, fn))\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "QBHki8fguCxU"
   },
   "source": [
    "# Particle Process Discriminator - Algorithm Selection\n",
    "\n",
    "This notebook explores several classifiers to discriminate between signal (events with 4 top quarks) and background (events with a top and antitop quark). Exploratory data analysis is performed in `Data Exporation and Preprocessing`, the reader is refered there for an analysis of the problem.  \n",
    "\n",
    "The dataset contains 100.000 samples, so for quick inference, ensembles of Decision Trees are considered. K-nearest Neighbours classifiers are too slow on inference.  \n",
    "\n",
    "Specifically, the following methods are considered:\n",
    "- Single Decision Tree\n",
    "- A boosted ensemble of Decision Trees\n",
    "- A bagged ensemble of Decision Trees\n",
    "- A Random Forest classifier\n",
    "\n",
    "Furthermore, a deep neural network is considered. This network can aliviate several limitations of Decision Trees, but requires more setup. A simple network is trained here, with more analysis provided in `Train Deep Neural Networks`.   \n",
    "\n",
    "Finally, the tabular dataset is converted to images, with the azimuthal angle and pseudorapidity decoding for the pixel location. A simple network is trained here, with an extensive analysis of symmetries provided in `Train Convolutional Neural Networks`. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "kYVAeSS8LbLV"
   },
   "source": [
    "# Compare sklearn classifiers with 5-fold cross validation\n",
    "The following classifiers are considered: \n",
    "- Single Decision Tree\n",
    "- A boosted ensemble of Decision Trees\n",
    "- A bagged ensemble of Decision Trees\n",
    "- A Random Forest classifier\n",
    "\n",
    "The bagged ensemble of Decision Trees is configured in four ways:\n",
    "- With Decision Trees of max. depth 4 or 8\n",
    "- With either 50% of data and features, or with 75% of data and features \n",
    "\n",
    "All ensemble classifiers are trained with 100 estimators. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "o41BpmhED75w"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Size of DataFrame: (100000, 248)\n"
     ]
    }
   ],
   "source": [
    "# read train and validation data\n",
    "data, features, target = load_data('data', trans='plain', norm='minmax')\n",
    "print(\"Size of DataFrame: {}\".format(data.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 136
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 3206571,
     "status": "ok",
     "timestamp": 1581258807364,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDmAFZb27lfHNgO3aNVuvnBzM0iEOwYGxP6Uq2t_A=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "rYZqLGTLxP-j",
    "outputId": "8ebdc12a-bce0-4d16-e161-737064679d1d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting Decision Tree..\n",
      "Starting AdaBoost, tiny DTs..\n",
      "Starting Bagging, small DTs, 50/50..\n",
      "Starting Bagging, medium DTs, 50/50..\n",
      "Starting Bagging, small DTs, 75/75..\n",
      "Starting Bagging, medium DTs, 75/75..\n",
      "Starting Random Forest, large DTs..\n",
      "Finished.\n"
     ]
    }
   ],
   "source": [
    "n_estimators = 100\n",
    "criterion = 'entropy'  # seems like the natural choice for physical processes\n",
    "n_jobs = -1  # number of processor cores to use (-1 to utilize all cores)\n",
    "verbose = 1  # verbosity (1 = verbose, 0=silent)\n",
    "\n",
    "classifier_options = [\n",
    "    DecisionTreeClassifier(criterion=criterion),\n",
    "    \n",
    "    AdaBoostClassifier(DecisionTreeClassifier(criterion=criterion, max_depth=2), \n",
    "                       n_estimators=n_estimators),\n",
    "    \n",
    "    BaggingClassifier(DecisionTreeClassifier(criterion=criterion, max_depth=4), \n",
    "                      n_estimators=n_estimators, max_samples=0.5, max_features=0.5, \n",
    "                      n_jobs=n_jobs, verbose=verbose),\n",
    "    \n",
    "    BaggingClassifier(DecisionTreeClassifier(criterion=criterion, max_depth=8), \n",
    "                      n_estimators=n_estimators, max_samples=0.5, max_features=0.5, \n",
    "                      n_jobs=n_jobs, verbose=verbose),\n",
    "    \n",
    "    BaggingClassifier(DecisionTreeClassifier(criterion=criterion, max_depth=4), \n",
    "                      n_estimators=n_estimators, max_samples=0.75, max_features=0.75, \n",
    "                      n_jobs=n_jobs, verbose=verbose),\n",
    "    \n",
    "    BaggingClassifier(DecisionTreeClassifier(criterion=criterion, max_depth=8), \n",
    "                      n_estimators=n_estimators, max_samples=0.75, max_features=0.75, \n",
    "                      n_jobs=n_jobs, verbose=verbose),\n",
    "    \n",
    "    RandomForestClassifier(criterion=criterion, n_estimators=n_estimators, \n",
    "                           n_jobs=n_jobs, verbose=verbose),\n",
    "]\n",
    "\n",
    "classifier_labels = ['Decision Tree', 'AdaBoost, tiny DTs', \n",
    "                     'Bagging, small DTs, 50/50', 'Bagging, medium DTs, 50/50', \n",
    "                     'Bagging, small DTs, 75/75', 'Bagging, medium DTs, 75/75', \n",
    "                     'Random Forest, large DTs']\n",
    "\n",
    "cv_folds = 5\n",
    "accuracies = np.zeros(shape=(len(classifier_options), cv_folds))\n",
    "\n",
    "for i, clf in enumerate(classifier_options):\n",
    "    print(\"Starting {}..\".format(classifier_labels[i]))\n",
    "    # fit classifier with cross validation\n",
    "    accuracies[i] = cross_val_score(clf, data[features].values, data[target], cv=cv_folds, n_jobs=n_jobs)\n",
    "print(\"Finished.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 356
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1157,
     "status": "ok",
     "timestamp": 1581269539104,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDmAFZb27lfHNgO3aNVuvnBzM0iEOwYGxP6Uq2t_A=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "ATL1lqd8lKyi",
    "outputId": "26f8d59a-be43-416f-86db-33ace3ea1961"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Baseline accuracy of Decision Tree: 0.90726 ± 0.00091\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "classifier_labels = ['Decision\\nTree', 'AdaBoost,\\nTiny', \n",
    "                     'Bagging,\\nSmall,\\n50/50', 'Bagging,\\nMedium,\\n50/50', \n",
    "                     'Bagging,\\nSmall,\\n75/75', 'Bagging,\\nMedium,\\n75/75', \n",
    "                     'Random\\nForest,\\nLarge']\n",
    "\n",
    "baseline = 0\n",
    "print(\"Baseline accuracy of {}: {:.5f} ± {:.5f}\".format(classifier_labels[baseline].replace('\\n', ' '),\n",
    "      np.mean(accuracies[baseline]), np.std(accuracies[baseline])))\n",
    "plot_accuracy_gain(accuracies, classifier_labels, baseline=0, fn=None, label_rotation=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Performance increases of ensemble over (unoptimized) Decision Tree is large.  \n",
    "**Show top performing classifiers**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of AdaBoost, Tiny: 0.93455 ± 0.00064\n",
      "Accuracy of Bagging, Medium, 50/50: 0.93536 ± 0.00101\n",
      "Accuracy of Bagging, Medium, 75/75: 0.93698 ± 0.00086\n",
      "Accuracy of Random Forest, Large: 0.93759 ± 0.00115\n"
     ]
    }
   ],
   "source": [
    "selection = [1, 3, 5, 6]\n",
    "labels = np.array(classifier_labels)[selection]\n",
    "plot_accuracy_gain(accuracies[selection], labels, baseline=0, fn=None, label_rotation=0)\n",
    "for i in selection:\n",
    "    print(\"Accuracy of {}: {:.5f} ± {:.5f}\".format(classifier_labels[i].replace('\\n', ' '),\n",
    "      np.mean(accuracies[i]), np.std(accuracies[i])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "snrvHRqlcTBt"
   },
   "source": [
    "## Number of estimators for Random Forest classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 4409,
     "status": "ok",
     "timestamp": 1581269832382,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDmAFZb27lfHNgO3aNVuvnBzM0iEOwYGxP6Uq2t_A=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "QKbhgwv4onFY",
    "outputId": "29b504d8-1faf-404f-8446-9650121c754e"
   },
   "outputs": [],
   "source": [
    "# read data\n",
    "data, features, target = load_data('data', trans='both')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 85
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 482822,
     "status": "ok",
     "timestamp": 1581029455581,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDmAFZb27lfHNgO3aNVuvnBzM0iEOwYGxP6Uq2t_A=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "i7ptpQkAp9NK",
    "outputId": "6369a9cb-c7da-4289-fb84-7c8dabcf2a6d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting 10 estimators..\n",
      "Starting 30 estimators..\n",
      "Starting 100 estimators..\n",
      "Starting 300 estimators..\n",
      "Starting 1000 estimators..\n",
      "Finished.\n"
     ]
    }
   ],
   "source": [
    "n_options = [10, 30, 100, 300, 1000]\n",
    "cv_folds = 5\n",
    "accuracies = np.zeros(shape=(len(n_options), cv_folds))\n",
    "n_jobs = -1  # number of processor cores to use (-1 to utilize all cores)\n",
    "verbose = 0  # verbosity of sklearn classifier (1 = verbose, 0=silent)\n",
    "\n",
    "for i, n_estimators in enumerate(n_options):\n",
    "    print(\"Starting {} estimators..\".format(n_estimators))\n",
    "    clf = RandomForestClassifier(n_estimators=n_estimators, verbose=verbose, n_jobs=n_jobs)\n",
    "    # fit RF classifier with cross validation\n",
    "    accuracies[i] = cross_val_score(clf, data[features].values, data[target], cv=cv_folds, \n",
    "                                    n_jobs=n_jobs, verbose=verbose)\n",
    "print(\"Finished.\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 323
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 774,
     "status": "ok",
     "timestamp": 1581030715946,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDmAFZb27lfHNgO3aNVuvnBzM0iEOwYGxP6Uq2t_A=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "AuQ4rjpYq8zn",
    "outputId": "da3ca517-728e-4313-a775-37939b6ad91a"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "labels = n_options\n",
    "plot_accuracy_gain(accuracies, labels, baseline=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "tBHjwCJPiGzQ"
   },
   "source": [
    "**Performance of Random Forest classifier with most Decision Trees**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 757,
     "status": "ok",
     "timestamp": 1581268083846,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mDmAFZb27lfHNgO3aNVuvnBzM0iEOwYGxP6Uq2t_A=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "NleHK5A260k-",
    "outputId": "cf40155d-22ef-4d2f-ae92-6f2e6dfa875f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RF accuracy: 0.93880 ± 0.00053\n"
     ]
    }
   ],
   "source": [
    "print(\"RF accuracy: {:.5f} ± {:.5f}\".format(np.mean(accuracies[-1]), np.std(accuracies[-1])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Deep Neural Network\n",
    "Train a multi layered perceptron. We will start with a simple network with 128, 64 and 32 hidden nodes. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "8dAIeLOvuCxY"
   },
   "source": [
    "## Read preprocessed data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-02-01T16:07:00.143259Z",
     "start_time": "2020-02-01T16:06:54.115047Z"
    },
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 6827,
     "status": "ok",
     "timestamp": 1584530972037,
     "user": {
      "displayName": "Joeran Bosma",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14GiEvCunvq33iGZ3RU-P9BMF7L60e8LoQ3Y1AfZJ4Q=s64",
      "userId": "05378465311079003812"
     },
     "user_tz": -60
    },
    "id": "TOpr4NETuCxa",
    "outputId": "dfb5204d-1221-4233-a143-2fbe9fed67a5"
   },
   "outputs": [],
   "source": [
    "# read train and validation data\n",
    "data, features, target = load_data('data', trans='plain', norm='minmax')\n",
    "train, validation = train_test_split(data, test_size=0.2, random_state=random_state)\n",
    "\n",
    "# convert to numpy arrays\n",
    "X_train, y_train = train[features].values, train[target].values\n",
    "X_val, y_val = validation[features].values, validation[target].values"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4_zipQxRWI5T"
   },
   "source": [
    "## Multi layered perceptron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "id                         DNN baseline\n",
      "architecture                           \n",
      "loss                binary_crossentropy\n",
      "activation                         relu\n",
      "final_activation                sigmoid\n",
      "dropout                               0\n",
      "epochs                               20\n",
      "batch_size                           32\n",
      "lr                                0.001\n",
      "lr_decay                              0\n",
      "optimizer                          adam\n",
      "norm                             minmax\n",
      "trans                             plain\n",
      "nodes1                              128\n",
      "nodes2                               64\n",
      "nodes3                               32\n",
      "nodes4                                0\n",
      "nodes5                                0\n",
      "dropout1                              0\n",
      "dropout2                              0\n",
      "dropout3                              0\n",
      "dropout4                              0\n",
      "dropout5                              0\n",
      "batch_norm                           []\n",
      "dtype: object\n"
     ]
    }
   ],
   "source": [
    "args = make_args(name='DNN baseline', wandb_init=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting fold 1/5 with 80000 train samples and 20000 validation samples\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 128)               17920     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 28,289\n",
      "Trainable params: 28,289\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting fold 2/5 with 80000 train samples and 20000 validation samples\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 128)               17920     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 28,289\n",
      "Trainable params: 28,289\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting fold 3/5 with 80000 train samples and 20000 validation samples\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 128)               17920     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 28,289\n",
      "Trainable params: 28,289\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting fold 4/5 with 80000 train samples and 20000 validation samples\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 128)               17920     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 28,289\n",
      "Trainable params: 28,289\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting fold 5/5 with 80000 train samples and 20000 validation samples\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 128)               17920     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 28,289\n",
      "Trainable params: 28,289\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cv_folds = 5\n",
    "dnn_accuracies = np.zeros(cv_folds)\n",
    "baseline_acc = np.mean(accuracies[-1]) # or None\n",
    "\n",
    "for i, cv_it in enumerate(range(1, 1+cv_folds)):\n",
    "    # perform cross validation splits\n",
    "    kf = KFold(n_splits=cv_folds, shuffle=True, random_state=random_state)\n",
    "    # bit wacky, but this selects the cv-th set of train/test indices (with cv starting at ONE)\n",
    "    for i in range(cv_it):\n",
    "        train_indices, val_indices = next(kf.split(data))\n",
    "    \n",
    "    # define train and validation sets\n",
    "    train, validation = data.iloc[train_indices], data.iloc[val_indices]\n",
    "    print(\"Starting fold {}/{} with {} train samples and {} validation samples\".format(\n",
    "        cv_it, cv_folds, len(train_indices), len(val_indices)\n",
    "    ))\n",
    "    \n",
    "    model, history = train_net(args, X_train, y_train, X_val, y_val,\n",
    "                           keras_verbose=0, evaluate=False)\n",
    "    \n",
    "    # save accuracy\n",
    "    y_pred = model.predict(X_val)\n",
    "    y_pred = (y_pred > 0.5) # convert confidence to classification\n",
    "    dnn_accuracies[i] = accuracy_score(y_val, y_pred)\n",
    "    \n",
    "    # show accuracies\n",
    "    f, axes = plt.subplots(1, 1, figsize=(12, 5))\n",
    "    ax = axes\n",
    "    ax.plot(history.history['acc'], label='Train')\n",
    "    ax.plot(history.history['val_acc'], label='Test')\n",
    "    ax.set_title(\"Accuracy\"); ax.set_ylabel(\"Accuracy\")\n",
    "    ax.set_xlabel(\"Epoch\");\n",
    "\n",
    "    if baseline_acc is not None:\n",
    "        ax.axhline(baseline_acc, linestyle=':', color='k', label='Random Forest')  # mean Random Forest accuracy\n",
    "\n",
    "    ax.legend()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DNN accuracy: 0.92528 ± 0.00246\n"
     ]
    }
   ],
   "source": [
    "print(\"DNN accuracy: {:.5f} ± {:.5f}\".format(np.mean(dnn_accuracies), np.std(dnn_accuracies)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "JRu5HnZDbCv0"
   },
   "source": [
    "The neural network performs similar to the Random Forest classifier, albeit a bit worse. So far, the Random Forest classifier performs best, with a validation accuracy just short of 94%. The dense neural network shows good potential and can probably still improve on the final validation accuracy of about 93%.  \n",
    "\n",
    "Note that these results are only exploratory, as no optimization has been done. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "M_RPWbcFWO4f"
   },
   "source": [
    "## Convolutional network\n",
    "The dense network performed well, but a convolutional network can possible provide additional information. The information about the energy and type of object is probably best suited for a dense network, but a convolutional neural network can perhaps extract more information about the spatial relations.  \n",
    "\n",
    "The exploration of the CNN is done in `Train Convolutional Neural Networks`. "
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "ML4PPA - Final - Joeran Bosma - Algorithm Selection.ipynb",
   "provenance": [
    {
     "file_id": "1lMw1y4fnw_7L_bfKv-44EAcJu-RGmaYS",
     "timestamp": 1581263939189
    }
   ]
  },
  "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}