{"cells":[{"metadata":{},"cell_type":"markdown","source":"**CNN keras for digit recognition of hand written digit images of size 28x28 **"},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"import pandas as pd\nimport numpy as np\nimport seaborn as sns\nfrom matplotlib import pyplot as plt\n\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import confusion_matrix\nimport itertools\n\nfrom keras.utils.np_utils import to_categorical\nfrom keras.models import Sequential\nfrom keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D\nfrom keras.optimizers import Adam\nfrom keras.callbacks import ReduceLROnPlateau\n\nsns.set(style='white', context='notebook', palette='bright')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"> Data Preparation and EDA"},{"metadata":{"trusted":true},"cell_type":"code","source":"#load training data\ntrain_orig = pd.read_csv(\"../input/digit-recognizer/train.csv\")\ntrain_orig.describe()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#checking distribution of target variable\ndef check_label_distribution(target):\n    print(target.value_counts())\n    sns.countplot(target)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"check_label_distribution(train_orig[\"label\"])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#separate out target variable\ndef separate_x_y(data):\n    return (data.drop(\"label\",axis=1), data[\"label\"])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"x_train,y_train = separate_x_y(train_orig)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#normalize pixel values\ndef normalize_pixels(train):\n    return train/255.","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"x_train = normalize_pixels(x_train)\nx_train.describe()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#reshaping for feeding into CNN\ndef reshaping_images(images,height, width ,canal):\n    return images.values.reshape(-1, height, width, canal)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#reshaping \nx_train = reshaping_images(x_train, 28, 28, 1)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_train = to_categorical(y_train, num_classes=10)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"> train test split"},{"metadata":{"trusted":true},"cell_type":"code","source":"X_train, X_dev, Y_train, Y_dev = train_test_split(x_train, y_train, test_size=0.2, random_state = 7)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(X_train.shape)\nprint(Y_train.shape)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"plt.imshow(X_train[4][:,:,0])","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"> Model Building"},{"metadata":{"trusted":true},"cell_type":"code","source":"model = Sequential()\nmodel.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu', input_shape=(28,28,1)))\nmodel.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu', input_shape=(28,28,1)))\nmodel.add(MaxPool2D(pool_size=(2,2)))\nmodel.add(Dropout(0.5))\nmodel.add(Conv2D(filters=64, kernel_size=(3,3), activation='relu', input_shape=(28,28,1)))\nmodel.add(Conv2D(filters=64, kernel_size=(3,3), activation='relu', input_shape=(28,28,1)))\nmodel.add(MaxPool2D(pool_size=(2,2), strides=(2,2)))\nmodel.add(Dropout(0.25))\n\nmodel.add(Flatten())\nmodel.add(Dense(256, activation = \"relu\"))\nmodel.add(Dropout(0.5))\nmodel.add(Dense(10, activation = \"softmax\"))\n","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"> Adam Optimizer for loss function"},{"metadata":{"trusted":true},"cell_type":"code","source":"optimizer = Adam(learning_rate=0.001, beta_1=0.9, beta_2=0.999)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=[\"accuracy\"])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"learning_rate_reduction = ReduceLROnPlateau(monitor='val_accuracy', \n                                            patience=5, \n                                            verbose=1, \n                                            factor=0.2, \n                                            min_lr=0.0001)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"epoch= 31\nbatch_size= 64","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model_fit = model.fit(x=X_train, y=Y_train, batch_size=batch_size, epochs=epoch, verbose=1, callbacks=[learning_rate_reduction],validation_data=(X_dev, Y_dev), shuffle=True)\n","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"> We obtained Dev Set Accuracy = 99.35%"},{"metadata":{},"cell_type":"markdown","source":"> Confusion matrix"},{"metadata":{"trusted":true},"cell_type":"code","source":"Y_pred = model.predict(X_dev)\nconfusion_matrix(np.argmax(Y_dev,axis = 1) , np.argmax(Y_pred,axis = 1) )","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Can see confusion between 3-7 and 4-9"},{"metadata":{"trusted":true},"cell_type":"code","source":"arr=np.argmax(Y_dev,axis = 1)\npred_arr = np.argmax(Y_pred,axis = 1)\nind=np.where(arr==1)\nind2=np.where(pred_arr==7)\nmatch_ind = np.intersect1d(ind,ind2)\nif len(match_ind)>0:\n    match_ind=match_ind[0]\n    print(\"Actual: \",arr[match_ind])\n    print(\"Predicted: \",pred_arr[match_ind])\n    plt.imshow(X_dev[match_ind][:,:,0])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"arr=np.argmax(Y_dev,axis = 1)\npred_arr = np.argmax(Y_pred,axis = 1)\nind=np.where(arr==4)\nind2=np.where(pred_arr==9)\nmatch_ind = np.intersect1d(ind,ind2)\nif len(match_ind)>0:\n    match_ind=match_ind[0]\n    print(\"Actual: \",arr[match_ind])\n    print(\"Predicted: \",pred_arr[match_ind])\n    plt.imshow(X_dev[match_ind][:,:,0])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"arr=np.argmax(Y_dev,axis = 1)\npred_arr = np.argmax(Y_pred,axis = 1)\nind=np.where(arr==8)\nind2=np.where(pred_arr==9)\nmatch_ind = np.intersect1d(ind,ind2)\nif len(match_ind)>0:\n    match_ind=match_ind[0]\n    print(\"Actual: \",arr[match_ind])\n    print(\"Predicted: \",pred_arr[match_ind])\n    plt.imshow(X_dev[match_ind][:,:,0])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"arr=np.argmax(Y_dev,axis = 1)\npred_arr = np.argmax(Y_pred,axis = 1)\nind=np.where(arr==7)\nind2=np.where(pred_arr==2)\nmatch_ind = np.intersect1d(ind,ind2)\nif len(match_ind)>0:\n    match_ind=match_ind[0]\n    print(\"Actual: \",arr[match_ind])\n    print(\"Predicted: \",pred_arr[match_ind])\n    plt.imshow(X_dev[match_ind][:,:,0])","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"> Evaluating on test data"},{"metadata":{"trusted":true},"cell_type":"code","source":"test_orig = pd.read_csv(\"../input/digit-recognizer/test.csv\")\ntest = normalize_pixels(test_orig)\ntest = reshaping_images(test, 28, 28, 1)\nresults = model.predict(test)\nresults = np.argmax(results,axis = 1)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"results.shape\nresults = pd.Series(results,name=\"Label\")\nresults.describe()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submission = pd.concat([pd.Series(range(1,28001),name = \"ImageId\"),results],axis = 1)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submission","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submission.to_csv(\"cnn_adam.csv\",index=False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}