train-explained.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Transfer learning \n",
    "\n",
    "We start with two folders containing positive and negative samples for stadiums, due to the nature of the problem, classes are highly unbalanced. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we'll use class_weights to reinforce positive samples (minority class)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "stadium_images = os.listdir('stadiums/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "non_stadium_images = os.listdir('non_stadiums/')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "11972"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(non_stadium_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "47"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(stadium_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "stadium_df = pd.DataFrame()\n",
    "stadium_df['filename'] = stadium_images\n",
    "stadium_df['is_stadium'] = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "non_stadium_df = pd.DataFrame()\n",
    "non_stadium_df['filename'] = non_stadium_images\n",
    "non_stadium_df['is_stadium'] = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_df = pd.concat([non_stadium_df,stadium_df])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset_df = dataset_df.drop(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Extraction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll use resnet50 to extract image features from the positives and negative samples. Then use those features as input\n",
    "to a single layer network with dropout for classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's try out feature extraction for a single image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.applications import resnet50\n",
    "from keras.applications.resnet50 import preprocess_input\n",
    "from keras.preprocessing import image\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[[[ 0.08236177  0.05985396  0.08081593 ...,  0.16119762  0.81385475\n",
      "     0.42690596]]]]\n",
      "(1, 1, 1, 2048)\n"
     ]
    }
   ],
   "source": [
    "model = resnet50.ResNet50(weights='imagenet', include_top=False)\n",
    "#model.summary()\n",
    "\n",
    "img_path = 'stadiums/Estadio Alberto J. Armando.png'\n",
    "img = image.load_img(img_path, target_size=(256, 256))\n",
    "img_data = image.img_to_array(img)\n",
    "img_data = np.expand_dims(img_data, axis=0)\n",
    "img_data = preprocess_input(img_data)\n",
    "\n",
    "resnet_features = model.predict(img_data)\n",
    "print(resnet_features)\n",
    "print(resnet_features.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "an image is then represented by an array of 2048 floats taken from the output of the resnet. Let's extract \n",
    "features for the rest of the stadium images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "image_features = []\n",
    "for x in stadium_images:\n",
    "    if x.endswith('.png'):\n",
    "        img = image.load_img('stadiums/'+x, target_size=(256, 256))\n",
    "        img_data = image.img_to_array(img)\n",
    "        img_data = np.expand_dims(img_data, axis=0)\n",
    "        img_data = preprocess_input(img_data)\n",
    "        features = model.predict(img_data)\n",
    "        image_features.append(features.flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "stadium_features = image_features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "non_stadium_features = []\n",
    "for x in non_stadium_images:\n",
    "    if x.endswith('.jpg'):\n",
    "        img = image.load_img('non_stadiums//'+x, target_size=(256, 256))\n",
    "        img_data = image.img_to_array(img)\n",
    "        img_data = np.expand_dims(img_data, axis=0)\n",
    "        img_data = preprocess_input(img_data)\n",
    "        features = model.predict(img_data)\n",
    "        non_stadium_features.append(features.flatten())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we saved extracted features so that we don't have to calculate them again."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "#save for future usage\n",
    "np.array(non_stadium_features).dump(open('non_stadium_features.pickle','wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "non_stadium_features = np.load(open('non_stadium_features.pickle','rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# simple NN to train with static features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#  training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We connect the outputs of the resnet50  with a dense single layer network. We add dropout to avoid overfitting. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model():\n",
    "    from keras import models\n",
    "    from keras import layers\n",
    "    from keras import optimizers\n",
    "    model = models.Sequential()\n",
    "    model.add(layers.Dense(256, activation='relu', input_dim=2048))\n",
    "    model.add(layers.Dropout(0.5))\n",
    "    model.add(layers.Dense(1, activation='sigmoid'))\n",
    "    model.compile(optimizer=optimizers.RMSprop(lr=2e-5),\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['acc'])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "45"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(stadium_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "12015"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import pandas as pd\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.utils import class_weight\n",
    "\n",
    "    \n",
    "\n",
    "def train(model,stadium_features,non_stadium_features):\n",
    "    features = np.concatenate((np.array(stadium_features),np.array(non_stadium_features)))\n",
    "    labels = np.concatenate((np.full(len(stadium_features),1),np.zeros(len(non_stadium_features))))\n",
    "    \n",
    "\n",
    "\n",
    "    train_features,val_features,train_labels,val_labels = train_test_split(features,labels,test_size=0.3)\n",
    "    print(\"positive samples on training set %s\" % sum(train_labels))\n",
    "    print(\"positive samples on validation set %s\" % sum(val_labels))\n",
    "    class_weights = class_weight.compute_class_weight('balanced',\n",
    "                                                 np.unique(train_labels),\n",
    "                                                 train_labels)\n",
    "    \n",
    "    print(\" class weights: %s\" % class_weights)\n",
    "    print(class_weights)\n",
    "    history = model.fit(x=train_features,y=train_labels,\n",
    "                    epochs=30,\n",
    "                    batch_size=20,validation_data=(val_features,val_labels),\n",
    "                    class_weight=class_weights,verbose=2)\n",
    "    \n",
    "    predictions = model.predict_proba(val_features)\n",
    "    results = list(map(lambda x,y: (x,y),predictions,val_labels))\n",
    "    results_df = pd.DataFrame.from_records(results)\n",
    "    results_df.columns=['prediction','label']\n",
    "    cm = confusion_matrix(y_true = results_df['label'],y_pred=results_df['prediction']>0.4,labels=[0.0,1.0])\n",
    "    print(cm)\n",
    "    report = classification_report(y_true = results_df['label'],y_pred=results_df['prediction']>0.3,labels=[0.0,1.0])\n",
    "    print(report)\n",
    "    \n",
    "    return history\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    " I'll try out training 3 different models, each of them involving 1k, 3k and 5k negative samples amongst the 45 positive samples. We then will test if it has any effect on the precision / accuracy for both classes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "positive samples on training set 30.0\n",
      "positive samples on validation set 15.0\n",
      " class weights: [  0.50428449  58.85      ]\n",
      "[  0.50428449  58.85      ]\n",
      "Train on 3531 samples, validate on 1514 samples\n",
      "Epoch 1/30\n",
      " - 2s - loss: 0.0817 - acc: 0.9881 - val_loss: 0.0720 - val_acc: 0.9901\n",
      "Epoch 2/30\n",
      " - 1s - loss: 0.0508 - acc: 0.9915 - val_loss: 0.0609 - val_acc: 0.9901\n",
      "Epoch 3/30\n",
      " - 1s - loss: 0.0447 - acc: 0.9915 - val_loss: 0.0475 - val_acc: 0.9901\n",
      "Epoch 4/30\n",
      " - 1s - loss: 0.0361 - acc: 0.9915 - val_loss: 0.0382 - val_acc: 0.9901\n",
      "Epoch 5/30\n",
      " - 1s - loss: 0.0280 - acc: 0.9918 - val_loss: 0.0342 - val_acc: 0.9901\n",
      "Epoch 6/30\n",
      " - 1s - loss: 0.0258 - acc: 0.9921 - val_loss: 0.0303 - val_acc: 0.9901\n",
      "Epoch 7/30\n",
      " - 1s - loss: 0.0210 - acc: 0.9929 - val_loss: 0.0245 - val_acc: 0.9908\n",
      "Epoch 8/30\n",
      " - 1s - loss: 0.0175 - acc: 0.9946 - val_loss: 0.0264 - val_acc: 0.9908\n",
      "Epoch 9/30\n",
      " - 1s - loss: 0.0153 - acc: 0.9946 - val_loss: 0.0261 - val_acc: 0.9908\n",
      "Epoch 10/30\n",
      " - 1s - loss: 0.0158 - acc: 0.9938 - val_loss: 0.0198 - val_acc: 0.9927\n",
      "Epoch 11/30\n",
      " - 1s - loss: 0.0135 - acc: 0.9946 - val_loss: 0.0210 - val_acc: 0.9921\n",
      "Epoch 12/30\n",
      " - 1s - loss: 0.0134 - acc: 0.9960 - val_loss: 0.0207 - val_acc: 0.9921\n",
      "Epoch 13/30\n",
      " - 1s - loss: 0.0110 - acc: 0.9960 - val_loss: 0.0192 - val_acc: 0.9941\n",
      "Epoch 14/30\n",
      " - 1s - loss: 0.0118 - acc: 0.9955 - val_loss: 0.0181 - val_acc: 0.9947\n",
      "Epoch 15/30\n",
      " - 1s - loss: 0.0088 - acc: 0.9977 - val_loss: 0.0167 - val_acc: 0.9947\n",
      "Epoch 16/30\n",
      " - 1s - loss: 0.0098 - acc: 0.9969 - val_loss: 0.0167 - val_acc: 0.9947\n",
      "Epoch 17/30\n",
      " - 1s - loss: 0.0118 - acc: 0.9958 - val_loss: 0.0151 - val_acc: 0.9947\n",
      "Epoch 18/30\n",
      " - 1s - loss: 0.0065 - acc: 0.9972 - val_loss: 0.0153 - val_acc: 0.9947\n",
      "Epoch 19/30\n",
      " - 1s - loss: 0.0068 - acc: 0.9980 - val_loss: 0.0160 - val_acc: 0.9947\n",
      "Epoch 20/30\n",
      " - 1s - loss: 0.0085 - acc: 0.9963 - val_loss: 0.0130 - val_acc: 0.9967\n",
      "Epoch 21/30\n",
      " - 1s - loss: 0.0071 - acc: 0.9977 - val_loss: 0.0152 - val_acc: 0.9960\n",
      "Epoch 22/30\n",
      " - 1s - loss: 0.0082 - acc: 0.9969 - val_loss: 0.0142 - val_acc: 0.9967\n",
      "Epoch 23/30\n",
      " - 1s - loss: 0.0067 - acc: 0.9972 - val_loss: 0.0132 - val_acc: 0.9974\n",
      "Epoch 24/30\n",
      " - 1s - loss: 0.0055 - acc: 0.9980 - val_loss: 0.0151 - val_acc: 0.9960\n",
      "Epoch 25/30\n",
      " - 1s - loss: 0.0060 - acc: 0.9977 - val_loss: 0.0123 - val_acc: 0.9980\n",
      "Epoch 26/30\n",
      " - 1s - loss: 0.0049 - acc: 0.9983 - val_loss: 0.0129 - val_acc: 0.9974\n",
      "Epoch 27/30\n",
      " - 1s - loss: 0.0043 - acc: 0.9989 - val_loss: 0.0132 - val_acc: 0.9974\n",
      "Epoch 28/30\n",
      " - 1s - loss: 0.0041 - acc: 0.9986 - val_loss: 0.0129 - val_acc: 0.9974\n",
      "Epoch 29/30\n",
      " - 1s - loss: 0.0037 - acc: 0.9986 - val_loss: 0.0113 - val_acc: 0.9980\n",
      "Epoch 30/30\n",
      " - 1s - loss: 0.0050 - acc: 0.9992 - val_loss: 0.0122 - val_acc: 0.9980\n",
      "[[1499    0]\n",
      " [   3   12]]\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "        0.0       1.00      1.00      1.00      1499\n",
      "        1.0       1.00      0.80      0.89        15\n",
      "\n",
      "avg / total       1.00      1.00      1.00      1514\n",
      "\n",
      "positive samples on training set 32.0\n",
      "positive samples on validation set 13.0\n",
      " class weights: [  0.52288984  11.421875  ]\n",
      "[  0.52288984  11.421875  ]\n",
      "Train on 731 samples, validate on 314 samples\n",
      "Epoch 1/30\n",
      " - 2s - loss: 0.6283 - acc: 0.7278 - val_loss: 0.1980 - val_acc: 0.9586\n",
      "Epoch 2/30\n",
      " - 0s - loss: 0.2157 - acc: 0.9425 - val_loss: 0.1713 - val_acc: 0.9586\n",
      "Epoch 3/30\n",
      " - 0s - loss: 0.1826 - acc: 0.9480 - val_loss: 0.1458 - val_acc: 0.9586\n",
      "Epoch 4/30\n",
      " - 0s - loss: 0.1292 - acc: 0.9590 - val_loss: 0.1178 - val_acc: 0.9586\n",
      "Epoch 5/30\n",
      " - 0s - loss: 0.1117 - acc: 0.9631 - val_loss: 0.0979 - val_acc: 0.9586\n",
      "Epoch 6/30\n",
      " - 0s - loss: 0.0677 - acc: 0.9726 - val_loss: 0.0912 - val_acc: 0.9618\n",
      "Epoch 7/30\n",
      " - 0s - loss: 0.0757 - acc: 0.9713 - val_loss: 0.0776 - val_acc: 0.9650\n",
      "Epoch 8/30\n",
      " - 0s - loss: 0.0553 - acc: 0.9754 - val_loss: 0.0638 - val_acc: 0.9682\n",
      "Epoch 9/30\n",
      " - 0s - loss: 0.0622 - acc: 0.9754 - val_loss: 0.0551 - val_acc: 0.9841\n",
      "Epoch 10/30\n",
      " - 0s - loss: 0.0579 - acc: 0.9795 - val_loss: 0.0504 - val_acc: 0.9841\n",
      "Epoch 11/30\n",
      " - 0s - loss: 0.0504 - acc: 0.9781 - val_loss: 0.0468 - val_acc: 0.9841\n",
      "Epoch 12/30\n",
      " - 0s - loss: 0.0490 - acc: 0.9808 - val_loss: 0.0430 - val_acc: 0.9841\n",
      "Epoch 13/30\n",
      " - 0s - loss: 0.0367 - acc: 0.9877 - val_loss: 0.0443 - val_acc: 0.9841\n",
      "Epoch 14/30\n",
      " - 0s - loss: 0.0318 - acc: 0.9891 - val_loss: 0.0432 - val_acc: 0.9841\n",
      "Epoch 15/30\n",
      " - 0s - loss: 0.0244 - acc: 0.9932 - val_loss: 0.0395 - val_acc: 0.9841\n",
      "Epoch 16/30\n",
      " - 0s - loss: 0.0310 - acc: 0.9891 - val_loss: 0.0368 - val_acc: 0.9841\n",
      "Epoch 17/30\n",
      " - 0s - loss: 0.0244 - acc: 0.9945 - val_loss: 0.0362 - val_acc: 0.9841\n",
      "Epoch 18/30\n",
      " - 0s - loss: 0.0241 - acc: 0.9904 - val_loss: 0.0372 - val_acc: 0.9841\n",
      "Epoch 19/30\n",
      " - 0s - loss: 0.0252 - acc: 0.9918 - val_loss: 0.0346 - val_acc: 0.9841\n",
      "Epoch 20/30\n",
      " - 0s - loss: 0.0166 - acc: 0.9959 - val_loss: 0.0319 - val_acc: 0.9841\n",
      "Epoch 21/30\n",
      " - 0s - loss: 0.0174 - acc: 0.9945 - val_loss: 0.0344 - val_acc: 0.9841\n",
      "Epoch 22/30\n",
      " - 0s - loss: 0.0170 - acc: 0.9973 - val_loss: 0.0286 - val_acc: 0.9841\n",
      "Epoch 23/30\n",
      " - 0s - loss: 0.0162 - acc: 0.9932 - val_loss: 0.0290 - val_acc: 0.9841\n",
      "Epoch 24/30\n",
      " - 0s - loss: 0.0159 - acc: 0.9959 - val_loss: 0.0261 - val_acc: 0.9841\n",
      "Epoch 25/30\n",
      " - 0s - loss: 0.0133 - acc: 0.9986 - val_loss: 0.0280 - val_acc: 0.9841\n",
      "Epoch 26/30\n",
      " - 0s - loss: 0.0137 - acc: 0.9973 - val_loss: 0.0253 - val_acc: 0.9841\n",
      "Epoch 27/30\n",
      " - 0s - loss: 0.0147 - acc: 0.9959 - val_loss: 0.0239 - val_acc: 0.9904\n",
      "Epoch 28/30\n",
      " - 0s - loss: 0.0116 - acc: 0.9959 - val_loss: 0.0248 - val_acc: 0.9904\n",
      "Epoch 29/30\n",
      " - 0s - loss: 0.0114 - acc: 0.9959 - val_loss: 0.0277 - val_acc: 0.9841\n",
      "Epoch 30/30\n",
      " - 0s - loss: 0.0135 - acc: 0.9959 - val_loss: 0.0272 - val_acc: 0.9841\n",
      "[[301   0]\n",
      " [  3  10]]\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "        0.0       0.99      1.00      1.00       301\n",
      "        1.0       1.00      0.77      0.87        13\n",
      "\n",
      "avg / total       0.99      0.99      0.99       314\n",
      "\n",
      "positive samples on training set 29.0\n",
      "positive samples on validation set 16.0\n",
      " class weights: [  0.50689819  36.74137931]\n",
      "[  0.50689819  36.74137931]\n",
      "Train on 2131 samples, validate on 914 samples\n",
      "Epoch 1/30\n",
      " - 2s - loss: 0.2811 - acc: 0.8930 - val_loss: 0.1142 - val_acc: 0.9825\n",
      "Epoch 2/30\n",
      " - 0s - loss: 0.0874 - acc: 0.9864 - val_loss: 0.1005 - val_acc: 0.9825\n",
      "Epoch 3/30\n",
      " - 0s - loss: 0.0708 - acc: 0.9864 - val_loss: 0.0806 - val_acc: 0.9825\n",
      "Epoch 4/30\n",
      " - 0s - loss: 0.0609 - acc: 0.9864 - val_loss: 0.0627 - val_acc: 0.9825\n",
      "Epoch 5/30\n",
      " - 0s - loss: 0.0464 - acc: 0.9873 - val_loss: 0.0552 - val_acc: 0.9825\n",
      "Epoch 6/30\n",
      " - 0s - loss: 0.0413 - acc: 0.9878 - val_loss: 0.0497 - val_acc: 0.9825\n",
      "Epoch 7/30\n",
      " - 0s - loss: 0.0286 - acc: 0.9906 - val_loss: 0.0427 - val_acc: 0.9825\n",
      "Epoch 8/30\n",
      " - 0s - loss: 0.0295 - acc: 0.9901 - val_loss: 0.0363 - val_acc: 0.9847\n",
      "Epoch 9/30\n",
      " - 0s - loss: 0.0278 - acc: 0.9901 - val_loss: 0.0330 - val_acc: 0.9858\n",
      "Epoch 10/30\n",
      " - 0s - loss: 0.0256 - acc: 0.9916 - val_loss: 0.0332 - val_acc: 0.9858\n",
      "Epoch 11/30\n",
      " - 0s - loss: 0.0229 - acc: 0.9916 - val_loss: 0.0378 - val_acc: 0.9847\n",
      "Epoch 12/30\n",
      " - 0s - loss: 0.0199 - acc: 0.9906 - val_loss: 0.0318 - val_acc: 0.9858\n",
      "Epoch 13/30\n",
      " - 0s - loss: 0.0176 - acc: 0.9920 - val_loss: 0.0295 - val_acc: 0.9858\n",
      "Epoch 14/30\n",
      " - 0s - loss: 0.0183 - acc: 0.9930 - val_loss: 0.0236 - val_acc: 0.9869\n",
      "Epoch 15/30\n",
      " - 0s - loss: 0.0165 - acc: 0.9934 - val_loss: 0.0242 - val_acc: 0.9880\n",
      "Epoch 16/30\n",
      " - 0s - loss: 0.0153 - acc: 0.9944 - val_loss: 0.0232 - val_acc: 0.9880\n",
      "Epoch 17/30\n",
      " - 0s - loss: 0.0157 - acc: 0.9930 - val_loss: 0.0220 - val_acc: 0.9880\n",
      "Epoch 18/30\n",
      " - 0s - loss: 0.0119 - acc: 0.9953 - val_loss: 0.0198 - val_acc: 0.9902\n",
      "Epoch 19/30\n",
      " - 0s - loss: 0.0108 - acc: 0.9958 - val_loss: 0.0213 - val_acc: 0.9880\n",
      "Epoch 20/30\n",
      " - 0s - loss: 0.0133 - acc: 0.9934 - val_loss: 0.0163 - val_acc: 0.9934\n",
      "Epoch 21/30\n",
      " - 0s - loss: 0.0106 - acc: 0.9972 - val_loss: 0.0176 - val_acc: 0.9934\n",
      "Epoch 22/30\n",
      " - 0s - loss: 0.0116 - acc: 0.9962 - val_loss: 0.0183 - val_acc: 0.9912\n",
      "Epoch 23/30\n",
      " - 0s - loss: 0.0104 - acc: 0.9962 - val_loss: 0.0165 - val_acc: 0.9934\n",
      "Epoch 24/30\n",
      " - 0s - loss: 0.0066 - acc: 0.9986 - val_loss: 0.0182 - val_acc: 0.9923\n",
      "Epoch 25/30\n",
      " - 0s - loss: 0.0091 - acc: 0.9962 - val_loss: 0.0181 - val_acc: 0.9934\n",
      "Epoch 26/30\n",
      " - 0s - loss: 0.0068 - acc: 0.9977 - val_loss: 0.0194 - val_acc: 0.9923\n",
      "Epoch 27/30\n",
      " - 0s - loss: 0.0093 - acc: 0.9958 - val_loss: 0.0160 - val_acc: 0.9934\n",
      "Epoch 28/30\n",
      " - 0s - loss: 0.0079 - acc: 0.9967 - val_loss: 0.0182 - val_acc: 0.9923\n",
      "Epoch 29/30\n",
      " - 0s - loss: 0.0063 - acc: 0.9977 - val_loss: 0.0175 - val_acc: 0.9934\n",
      "Epoch 30/30\n",
      " - 0s - loss: 0.0069 - acc: 0.9981 - val_loss: 0.0154 - val_acc: 0.9934\n",
      "[[898   0]\n",
      " [  3  13]]\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "        0.0       1.00      1.00      1.00       898\n",
      "        1.0       1.00      0.88      0.93        16\n",
      "\n",
      "avg / total       1.00      1.00      1.00       914\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model_5000=build_model()\n",
    "history_5000= train(model_5000,stadium_features=stadium_features,non_stadium_features=non_stadium_features[np.random.choice(11970,size=5000)])\n",
    "\n",
    "model_1000=build_model()\n",
    "history_1000= train(model_1000,stadium_features=stadium_features,non_stadium_features=non_stadium_features[np.random.choice(11970,size=1000)])\n",
    "\n",
    "model_3000=build_model()\n",
    "history_3000= train(model_3000,stadium_features=stadium_features,non_stadium_features=non_stadium_features[np.random.choice(11970,size=3000)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "def plot_history(history,title=None):\n",
    "    # Get training and test loss histories\n",
    "    training_loss = history.history['loss']\n",
    "    test_loss = history.history['val_loss']\n",
    "\n",
    "    # Create count of the number of epochs\n",
    "    epoch_count = range(1, len(training_loss) + 1)\n",
    "\n",
    "    # Visualize loss history\n",
    "    plt.plot(epoch_count, training_loss, 'r--')\n",
    "    plt.plot(epoch_count, test_loss, 'b-')\n",
    "\n",
    "    plt.legend(['Training Loss', 'Test Loss'])\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.ylabel('Loss')\n",
    "    if title:\n",
    "        plt.title(title)\n",
    "    plt.show();\n",
    "\n",
    "    # Get training and test loss histories\n",
    "    training_loss = history.history['acc']\n",
    "    test_loss = history.history['val_acc']\n",
    "    plt.plot(epoch_count, training_loss, 'r--')\n",
    "    plt.plot(epoch_count, test_loss, 'b-')\n",
    "    plt.legend(['Training Acc', 'Val acc'])\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.ylabel('Loss')\n",
    "    if title:\n",
    "        plt.title(title)\n",
    "    plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_history(history_1000,title=\"1k negative samples\")\n",
    "plot_history(history_3000,title=\"3k negative samples\")\n",
    "plot_history(history_5000,title=\"5k negative samples\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "we see all three models behave in a similar fashion: \n",
    "- Learn to predict negative samples almost perfectly ( easiest).\n",
    "- Make some errors predicting positive samples. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We picked the model trained with 3000 negative samples as it renders the best f1_score. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_3000.save('resnet+dense_3000.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}