{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load in \n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\nimport cv2\n# Input data files are available in the \"../input/\" directory.\n# For example, running this (by clicking run or pressing Shift+Enter) will list the files in the input directory\n\nimport os\nprint(os.listdir(\"../input\"))\n\n# Any results you write to the current directory are saved as output.","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"data = os.listdir(\"../input/train_images\")\ndata_labels = pd.read_csv('../input/train.csv')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"x = data_labels['id_code']\ny = data_labels['diagnosis']","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_img = []\ny_p = []\ndef create_training_set(label, path):\n    img = cv2.imread(path, cv2.IMREAD_COLOR)\n    img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n    img = cv2.resize(img, (32,32))\n    X_img.append(np.array(img))\n    y_p.append(str(label))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from tqdm import tqdm\nTRAIN_DIR = '../input/train_images'\nfor id_code, diagnosis in tqdm(zip(x,y)):\n    path = os.path.join(TRAIN_DIR, '{}.png'.format(id_code))\n    create_training_set(diagnosis, path)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from keras.utils import to_categorical\n#Y = to_categorical(y_p)\nY = np.array(y_p)\nX= np.array(X_img)\nX=X/255","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Y = np.array(y_p)\nY = Y.astype(int)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.model_selection import train_test_split\nX_train, X_valid, Y_train, Y_valid = train_test_split(X, Y, test_size=0.20, random_state=0)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_train = X_train.reshape(2929,1024)\nX_test = X_valid.reshape(733,1024)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.linear_model import LogisticRegression\nlr = LogisticRegression()\nlr.fit(X_train,Y_train)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_pred = lr.predict(X_test)\nfrom sklearn.metrics import accuracy_score as ac\nac(Y_valid, y_pred)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"markdown","source":"from keras.preprocessing.image import ImageDataGenerator\ntransform = ImageDataGenerator(\n    featurewise_center=True,\n    featurewise_std_normalization=True,\n    rotation_range=20,\n    width_shift_range=0.2,\n    height_shift_range=0.2,\n    horizontal_flip=True)\ntransform.fit(X_train)"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"transform.fit(X_valid)"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"from keras.layers import Dense,Convolution2D,Flatten,MaxPooling2D,Activation,Dropout\nfrom keras.models import Sequential\nfrom keras.layers.normalization import BatchNormalization"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"model = Sequential()\nmodel.add(Convolution2D(filters=16, kernel_size=2, padding='same', activation='relu', input_shape=(150,150,3)))\nmodel.add(Activation('elu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(BatchNormalization())\n\nmodel.add(Convolution2D(32, (3, 3), padding='same'))\nmodel.add(Activation('elu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(BatchNormalization())\n\nmodel.add(Convolution2D(64, (3, 3), padding='same'))\nmodel.add(Activation('relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(BatchNormalization())\n\nmodel.add(Flatten())\n\nmodel.add(Dense(128))\nmodel.add(Activation('relu'))\nmodel.add(Dropout(0.25))\nmodel.add(Dense(64))\nmodel.add(Activation('softmax'))\nmodel.add(Dropout(0.25))\nmodel.add(Dense(5))\nmodel.add(Activation('softmax'))\n\nmodel.summary()"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"model = Sequential()\nmodel.add(Convolution2D(3, (3, 3), padding='same',\n                 input_shape=(32,32,1)))\nmodel.add(Activation('relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(BatchNormalization())\n\n'''model.add(Convolution2D(32, (3, 3), padding='same'))\nmodel.add(Activation('relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(BatchNormalization()    '''\n\n'''model.add(Convolution2D(64, (3, 3), padding='same'))\nmodel.add(Activation('relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(BatchNormalization()) '''\n\nmodel.add(Convolution2D(64, (3, 3), padding='same'))\nmodel.add(Activation('relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(BatchNormalization())\n\nmodel.add(Flatten())\nmodel.add(Dense(128))\nmodel.add(Activation('relu'))\nmodel.add(Dropout(0.25))\nmodel.add(Dense(64))\nmodel.add(Activation('softmax'))\nmodel.add(Dense(5))\nmodel.add(Activation('softmax'))\n\nmodel.summary()"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"from keras import optimizers \nada = optimizers.Adagrad(lr = 0.0001)\nfrom keras import losses"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"model.compile(optimizer = ada, loss = losses.categorical_crossentropy, metrics = ['acc'])"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"model.fit(X_train,Y_train,epochs = 10,batch_size = 20)"},{"metadata":{"trusted":true},"cell_type":"markdown","source":"y_pred = model.predict(X_valid)\ny_pred = np.argmax(y_pred,axis = 1)\nnp.unique(y_pred)"},{"metadata":{},"cell_type":"markdown","source":"   This function calculates the Quadratic Kappa Metric. It returns the Quadratic Weighted Kappa metric score between    the actual and the predicted values of adoption rating."},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.metrics import confusion_matrix\ndef quadratic_kappa(actuals, preds, N):\n    w = np.zeros((N,N))\n    O = confusion_matrix(actuals, preds)\n    for i in range(len(w)): \n        for j in range(len(w)):\n            w[i][j] = float(((i-j)**2)/(N-1)**2)\n    \n    act_hist=np.zeros([N])\n    for item in actuals: \n        act_hist[item]+=1\n    \n    pred_hist=np.zeros([N])\n    for item in preds: \n        pred_hist[item]+=1\n                         \n    E = np.outer(act_hist, pred_hist);\n    E = E/E.sum();\n    O = O/O.sum();\n    \n    num=0\n    den=0\n    for i in range(len(w)):\n        for j in range(len(w)):\n            num+=w[i][j]*O[i][j]\n            den+=w[i][j]*E[i][j]\n    return (1 - (num/den))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"markdown","source":"Y_valid1 = np.argmax(Y_valid,axis =1)"},{"metadata":{"trusted":true},"cell_type":"code","source":"quadratic_kappa(Y_valid, y_pred,5)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_df = pd.read_csv('../input/test.csv')\ntest_df.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_ids = test_df['id_code']","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_images = []\ndef create_test_set(path):\n    img = cv2.imread(path,cv2.IMREAD_COLOR)\n    img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n    img = cv2.resize(img, (32,32))\n    test_images.append(np.array(img))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"for id_code in tqdm(test_ids):\n    path = os.path.join('../input/test_images','{}.png'.format(id_code))\n    create_test_set(path)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"markdown","source":"test_X=np.array(test_images)\ntransform.fit(test_X)\npred=model.predict(test_X)"},{"metadata":{"trusted":true},"cell_type":"code","source":"test_x = np.array(test_images)\ntest_x.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_x = test_x.reshape(1928,32*32)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"markdown","source":"pred = np.argmax(pred, axis=1)"},{"metadata":{"trusted":true},"cell_type":"code","source":"pred = lr.predict(test_x)\npred","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"np.unique(pred)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"markdown","source":"pred.astype(int)"},{"metadata":{"trusted":true},"cell_type":"code","source":"p = pd.DataFrame({'id_code':test_ids,'diagnosis':pred})\np.to_csv('submission.csv',index=False)","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.4","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}