ANN_First_Steps.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Getting Started With ANN"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading Data and Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('Churn_Modelling.csv') #using the churn dataset to predict wether a customer exits or stays\n",
    "X = data.iloc[:, 3:13].values\n",
    "Y = data.iloc[:,13].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import OneHotEncoder,LabelEncoder #Encoding\n",
    "labelencoder_X_1 = LabelEncoder()\n",
    "X[:,1] = labelencoder_X_1.fit_transform(X[:,1])\n",
    "labelencoder_X_2 = LabelEncoder()\n",
    "X[:,2] = labelencoder_X_2.fit_transform(X[:,2])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aayushi/anaconda3/lib/python3.7/site-packages/sklearn/preprocessing/_encoders.py:371: FutureWarning: The handling of integer data will change in version 0.22. Currently, the categories are determined based on the range [0, max(values)], while in the future they will be determined based on the unique values.\n",
      "If you want the future behaviour and silence this warning, you can specify \"categories='auto'\".\n",
      "In case you used a LabelEncoder before this OneHotEncoder to convert the categories to integers, then you can now use the OneHotEncoder directly.\n",
      "  warnings.warn(msg, FutureWarning)\n",
      "/home/aayushi/anaconda3/lib/python3.7/site-packages/sklearn/preprocessing/_encoders.py:392: DeprecationWarning: The 'categorical_features' keyword is deprecated in version 0.20 and will be removed in 0.22. You can use the ColumnTransformer instead.\n",
      "  \"use the ColumnTransformer instead.\", DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "onehotencoder = OneHotEncoder(categorical_features = [1]) #Telling the model not to rank the data\n",
    "X = onehotencoder.fit_transform(X).toarray()\n",
    "X = X[:, 1:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "X_train,X_test,Y_train,Y_test = train_test_split(X,Y,test_size = 0.2, random_state = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "sc = StandardScaler()\n",
    "X_train = sc.fit_transform(X_train)\n",
    "X_test = sc.transform(X_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Setting Up The Neural Network!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow import keras\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier = Sequential() \n",
    "classifier.add(Dense(units = 6,kernel_initializer = 'glorot_uniform', activation = 'relu',input_dim=11)) #Latest version of keras has 'units' instead of 'output_dim'\n",
    "classifier.add(Dense(units = 6, kernel_initializer = 'glorot_uniform',activation = 'relu',input_dim=11))\n",
    "classifier.add(Dense(units = 6, kernel_initializer = 'glorot_uniform',activation = 'relu',input_dim=11))\n",
    "classifier.add(Dense(units = 1,kernel_initializer='glorot_uniform',activation = 'sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier.compile(optimizer = 'adam',loss = 'binary_crossentropy',metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "W1108 01:58:41.075646 140068142126912 training.py:593] The `nb_epoch` argument in `fit` has been renamed `epochs`.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "8000/8000 [==============================] - 3s 404us/sample - loss: 0.5272 - acc: 0.7722\n",
      "Epoch 2/100\n",
      "8000/8000 [==============================] - 3s 388us/sample - loss: 0.4366 - acc: 0.8062\n",
      "Epoch 3/100\n",
      "8000/8000 [==============================] - 3s 378us/sample - loss: 0.4260 - acc: 0.8138\n",
      "Epoch 4/100\n",
      "8000/8000 [==============================] - 3s 412us/sample - loss: 0.4195 - acc: 0.8171\n",
      "Epoch 5/100\n",
      "8000/8000 [==============================] - 3s 386us/sample - loss: 0.4150 - acc: 0.8226\n",
      "Epoch 6/100\n",
      "8000/8000 [==============================] - 3s 363us/sample - loss: 0.4102 - acc: 0.8271\n",
      "Epoch 7/100\n",
      "8000/8000 [==============================] - 3s 396us/sample - loss: 0.4043 - acc: 0.8291\n",
      "Epoch 8/100\n",
      "8000/8000 [==============================] - 4s 446us/sample - loss: 0.3964 - acc: 0.8344\n",
      "Epoch 9/100\n",
      "8000/8000 [==============================] - ETA: 0s - loss: 0.3841 - acc: 0.840 - 4s 444us/sample - loss: 0.3851 - acc: 0.8397\n",
      "Epoch 10/100\n",
      "8000/8000 [==============================] - 4s 471us/sample - loss: 0.3715 - acc: 0.8472\n",
      "Epoch 11/100\n",
      "8000/8000 [==============================] - 4s 452us/sample - loss: 0.3593 - acc: 0.8530\n",
      "Epoch 12/100\n",
      "8000/8000 [==============================] - 4s 458us/sample - loss: 0.3508 - acc: 0.8568\n",
      "Epoch 13/100\n",
      "8000/8000 [==============================] - 4s 504us/sample - loss: 0.3462 - acc: 0.8586\n",
      "Epoch 14/100\n",
      "8000/8000 [==============================] - 4s 484us/sample - loss: 0.3437 - acc: 0.8589\n",
      "Epoch 15/100\n",
      "8000/8000 [==============================] - 4s 477us/sample - loss: 0.3417 - acc: 0.8571\n",
      "Epoch 16/100\n",
      "8000/8000 [==============================] - 4s 490us/sample - loss: 0.3409 - acc: 0.8615\n",
      "Epoch 17/100\n",
      "8000/8000 [==============================] - 4s 527us/sample - loss: 0.3397 - acc: 0.8601\n",
      "Epoch 18/100\n",
      "8000/8000 [==============================] - 4s 526us/sample - loss: 0.3387 - acc: 0.8614\n",
      "Epoch 19/100\n",
      "8000/8000 [==============================] - 4s 490us/sample - loss: 0.3387 - acc: 0.8629\n",
      "Epoch 20/100\n",
      "8000/8000 [==============================] - 4s 477us/sample - loss: 0.3385 - acc: 0.8596\n",
      "Epoch 21/100\n",
      "8000/8000 [==============================] - 4s 457us/sample - loss: 0.3371 - acc: 0.8609\n",
      "Epoch 22/100\n",
      "8000/8000 [==============================] - 5s 623us/sample - loss: 0.3372 - acc: 0.8606\n",
      "Epoch 23/100\n",
      "8000/8000 [==============================] - 4s 498us/sample - loss: 0.3368 - acc: 0.8616\n",
      "Epoch 24/100\n",
      "8000/8000 [==============================] - 4s 462us/sample - loss: 0.3367 - acc: 0.8629\n",
      "Epoch 25/100\n",
      "8000/8000 [==============================] - 4s 496us/sample - loss: 0.3364 - acc: 0.8609\n",
      "Epoch 26/100\n",
      "8000/8000 [==============================] - 4s 523us/sample - loss: 0.3357 - acc: 0.8635\n",
      "Epoch 27/100\n",
      "8000/8000 [==============================] - 4s 500us/sample - loss: 0.3361 - acc: 0.8604\n",
      "Epoch 28/100\n",
      "8000/8000 [==============================] - 4s 560us/sample - loss: 0.3358 - acc: 0.8619\n",
      "Epoch 29/100\n",
      "8000/8000 [==============================] - 4s 492us/sample - loss: 0.3352 - acc: 0.8616\n",
      "Epoch 30/100\n",
      "8000/8000 [==============================] - 4s 522us/sample - loss: 0.3355 - acc: 0.8612\n",
      "Epoch 31/100\n",
      "8000/8000 [==============================] - 4s 471us/sample - loss: 0.3352 - acc: 0.8641\n",
      "Epoch 32/100\n",
      "8000/8000 [==============================] - 4s 535us/sample - loss: 0.3349 - acc: 0.8629\n",
      "Epoch 33/100\n",
      "8000/8000 [==============================] - 4s 546us/sample - loss: 0.3345 - acc: 0.8631\n",
      "Epoch 34/100\n",
      "8000/8000 [==============================] - 4s 515us/sample - loss: 0.3340 - acc: 0.8618\n",
      "Epoch 35/100\n",
      "8000/8000 [==============================] - 4s 509us/sample - loss: 0.3346 - acc: 0.8635\n",
      "Epoch 36/100\n",
      "8000/8000 [==============================] - 4s 515us/sample - loss: 0.3343 - acc: 0.8651\n",
      "Epoch 37/100\n",
      "8000/8000 [==============================] - 5s 584us/sample - loss: 0.3334 - acc: 0.8622\n",
      "Epoch 38/100\n",
      "8000/8000 [==============================] - 5s 597us/sample - loss: 0.3340 - acc: 0.8630\n",
      "Epoch 39/100\n",
      "8000/8000 [==============================] - 4s 460us/sample - loss: 0.3334 - acc: 0.8616\n",
      "Epoch 40/100\n",
      "8000/8000 [==============================] - 3s 417us/sample - loss: 0.3334 - acc: 0.8631\n",
      "Epoch 41/100\n",
      "8000/8000 [==============================] - 3s 433us/sample - loss: 0.3333 - acc: 0.8621\n",
      "Epoch 42/100\n",
      "8000/8000 [==============================] - 4s 444us/sample - loss: 0.3329 - acc: 0.8615\n",
      "Epoch 43/100\n",
      "8000/8000 [==============================] - 4s 445us/sample - loss: 0.3338 - acc: 0.8619\n",
      "Epoch 44/100\n",
      "8000/8000 [==============================] - 4s 438us/sample - loss: 0.3328 - acc: 0.8645\n",
      "Epoch 45/100\n",
      "8000/8000 [==============================] - 4s 446us/sample - loss: 0.3338 - acc: 0.8645\n",
      "Epoch 46/100\n",
      "8000/8000 [==============================] - 3s 390us/sample - loss: 0.3328 - acc: 0.8610\n",
      "Epoch 47/100\n",
      "8000/8000 [==============================] - 4s 470us/sample - loss: 0.3332 - acc: 0.8640\n",
      "Epoch 48/100\n",
      "8000/8000 [==============================] - 4s 447us/sample - loss: 0.3331 - acc: 0.8626\n",
      "Epoch 49/100\n",
      "8000/8000 [==============================] - 4s 453us/sample - loss: 0.3323 - acc: 0.8620\n",
      "Epoch 50/100\n",
      "8000/8000 [==============================] - 4s 471us/sample - loss: 0.3326 - acc: 0.8643\n",
      "Epoch 51/100\n",
      "8000/8000 [==============================] - 4s 465us/sample - loss: 0.3321 - acc: 0.8631\n",
      "Epoch 52/100\n",
      "8000/8000 [==============================] - 4s 491us/sample - loss: 0.3322 - acc: 0.8651\n",
      "Epoch 53/100\n",
      "8000/8000 [==============================] - 4s 511us/sample - loss: 0.3323 - acc: 0.8641\n",
      "Epoch 54/100\n",
      "8000/8000 [==============================] - 4s 444us/sample - loss: 0.3321 - acc: 0.8651\n",
      "Epoch 55/100\n",
      "8000/8000 [==============================] - 4s 499us/sample - loss: 0.3319 - acc: 0.8634\n",
      "Epoch 56/100\n",
      "8000/8000 [==============================] - 4s 461us/sample - loss: 0.3321 - acc: 0.8636\n",
      "Epoch 57/100\n",
      "8000/8000 [==============================] - 4s 469us/sample - loss: 0.3328 - acc: 0.8624\n",
      "Epoch 58/100\n",
      "8000/8000 [==============================] - 4s 507us/sample - loss: 0.3316 - acc: 0.8614\n",
      "Epoch 59/100\n",
      "8000/8000 [==============================] - 4s 458us/sample - loss: 0.3315 - acc: 0.8643\n",
      "Epoch 60/100\n",
      "8000/8000 [==============================] - 4s 484us/sample - loss: 0.3316 - acc: 0.8630\n",
      "Epoch 61/100\n",
      "8000/8000 [==============================] - 4s 479us/sample - loss: 0.3315 - acc: 0.8637\n",
      "Epoch 62/100\n",
      "8000/8000 [==============================] - 4s 450us/sample - loss: 0.3316 - acc: 0.8637\n",
      "Epoch 63/100\n",
      "8000/8000 [==============================] - 4s 462us/sample - loss: 0.3316 - acc: 0.8626\n",
      "Epoch 64/100\n",
      "8000/8000 [==============================] - 4s 463us/sample - loss: 0.3316 - acc: 0.8652\n",
      "Epoch 65/100\n",
      "8000/8000 [==============================] - 4s 476us/sample - loss: 0.3312 - acc: 0.8652\n",
      "Epoch 66/100\n",
      "8000/8000 [==============================] - 4s 459us/sample - loss: 0.3315 - acc: 0.8625\n",
      "Epoch 67/100\n",
      "8000/8000 [==============================] - 4s 476us/sample - loss: 0.3302 - acc: 0.8639\n",
      "Epoch 68/100\n",
      "8000/8000 [==============================] - 4s 459us/sample - loss: 0.3312 - acc: 0.8652\n",
      "Epoch 69/100\n",
      "8000/8000 [==============================] - 4s 515us/sample - loss: 0.3305 - acc: 0.8654\n",
      "Epoch 70/100\n",
      "8000/8000 [==============================] - 4s 508us/sample - loss: 0.3304 - acc: 0.8655\n",
      "Epoch 71/100\n",
      "8000/8000 [==============================] - 4s 496us/sample - loss: 0.3308 - acc: 0.8627\n",
      "Epoch 72/100\n",
      "8000/8000 [==============================] - 4s 494us/sample - loss: 0.3305 - acc: 0.8644\n",
      "Epoch 73/100\n",
      "8000/8000 [==============================] - 5s 595us/sample - loss: 0.3308 - acc: 0.8634\n",
      "Epoch 74/100\n",
      "8000/8000 [==============================] - 4s 494us/sample - loss: 0.3303 - acc: 0.8640\n",
      "Epoch 75/100\n",
      "8000/8000 [==============================] - 4s 480us/sample - loss: 0.3303 - acc: 0.8648\n",
      "Epoch 76/100\n",
      "8000/8000 [==============================] - 4s 523us/sample - loss: 0.3304 - acc: 0.8649\n",
      "Epoch 77/100\n",
      "8000/8000 [==============================] - 4s 499us/sample - loss: 0.3308 - acc: 0.8636\n",
      "Epoch 78/100\n",
      "8000/8000 [==============================] - 4s 508us/sample - loss: 0.3304 - acc: 0.8633\n",
      "Epoch 79/100\n",
      "8000/8000 [==============================] - 4s 505us/sample - loss: 0.3299 - acc: 0.8658\n",
      "Epoch 80/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8000/8000 [==============================] - 4s 464us/sample - loss: 0.3299 - acc: 0.8639\n",
      "Epoch 81/100\n",
      "8000/8000 [==============================] - 4s 439us/sample - loss: 0.3302 - acc: 0.8659\n",
      "Epoch 82/100\n",
      "8000/8000 [==============================] - 4s 472us/sample - loss: 0.3306 - acc: 0.8637\n",
      "Epoch 83/100\n",
      "8000/8000 [==============================] - 4s 459us/sample - loss: 0.3302 - acc: 0.8656\n",
      "Epoch 84/100\n",
      "8000/8000 [==============================] - 4s 445us/sample - loss: 0.3299 - acc: 0.8658\n",
      "Epoch 85/100\n",
      "8000/8000 [==============================] - 4s 487us/sample - loss: 0.3303 - acc: 0.8631\n",
      "Epoch 86/100\n",
      "8000/8000 [==============================] - 4s 532us/sample - loss: 0.3296 - acc: 0.8656\n",
      "Epoch 87/100\n",
      "8000/8000 [==============================] - 4s 444us/sample - loss: 0.3300 - acc: 0.8654\n",
      "Epoch 88/100\n",
      "8000/8000 [==============================] - 4s 445us/sample - loss: 0.3299 - acc: 0.8654\n",
      "Epoch 89/100\n",
      "8000/8000 [==============================] - 4s 480us/sample - loss: 0.3304 - acc: 0.8637\n",
      "Epoch 90/100\n",
      "8000/8000 [==============================] - 4s 487us/sample - loss: 0.3303 - acc: 0.8637\n",
      "Epoch 91/100\n",
      "8000/8000 [==============================] - 4s 487us/sample - loss: 0.3304 - acc: 0.8625\n",
      "Epoch 92/100\n",
      "8000/8000 [==============================] - 4s 467us/sample - loss: 0.3302 - acc: 0.8627\n",
      "Epoch 93/100\n",
      "8000/8000 [==============================] - 3s 431us/sample - loss: 0.3298 - acc: 0.8644\n",
      "Epoch 94/100\n",
      "8000/8000 [==============================] - 4s 504us/sample - loss: 0.3302 - acc: 0.8634\n",
      "Epoch 95/100\n",
      "8000/8000 [==============================] - 4s 485us/sample - loss: 0.3298 - acc: 0.8633\n",
      "Epoch 96/100\n",
      "8000/8000 [==============================] - 4s 557us/sample - loss: 0.3294 - acc: 0.8643\n",
      "Epoch 97/100\n",
      "8000/8000 [==============================] - 4s 458us/sample - loss: 0.3298 - acc: 0.8648\n",
      "Epoch 98/100\n",
      "8000/8000 [==============================] - 4s 507us/sample - loss: 0.3293 - acc: 0.8637\n",
      "Epoch 99/100\n",
      "8000/8000 [==============================] - 4s 454us/sample - loss: 0.3300 - acc: 0.8629\n",
      "Epoch 100/100\n",
      "8000/8000 [==============================] - 4s 460us/sample - loss: 0.3287 - acc: 0.8646\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f62d739bef0>"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classifier.fit(X_train, Y_train, batch_size = 10, nb_epoch = 100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y_pred = classifier.predict(X_test)\n",
    "Y_pred = (Y_pred>0.5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1512,   83],\n",
       "       [ 199,  206]])"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "cm = confusion_matrix(Y_test,Y_pred)\n",
    "cm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plotting Accuracy and Loss for the Test Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Value')"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(classifier.history.history['acc'])\n",
    "plt.plot(classifier.history.history['loss'])\n",
    "plt.legend(['Accuracy', 'Loss'], loc='upper left')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Value')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next Steps : Tune the hyper-parameters to get the highest possible accuracy."
   ]
  },
  {
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}