# This Python 3 environment comes with many helpful analytics libraries installed
# It is defined by the kaggle/python docker image: https://github.com/kaggle/docker-python
# For example, here's several helpful packages to load in 

import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)

# Input data files are available in the "../input/" directory.
# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory

import os
for dirname, _, filenames in os.walk('/kaggle/input'):
    for filename in filenames:
        print(os.path.join(dirname, filename))

# Any results you write to the current directory are saved as output.

import matplotlib.pyplot as plt
import matplotlib.image as maping
import seaborn as sns

np.random.seed(2)

from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix
import itertools

from keras.utils.np_utils import to_categorical # for one hot encoding
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D
from keras.optimizers import RMSprop
from keras.preprocessing.image import ImageDataGenerator
from keras.callbacks import ReduceLROnPlateau #Reduce learning rate when a metric has stopped improving.
from keras.wrappers.scikit_learn import KerasClassifier
from sklearn.model_selection import GridSearchCV

sns.set(style='white', context='notebook', palette='deep')

mnist_train = pd.read_csv("../input/Kannada-MNIST/train.csv")

mnist_test = pd.read_csv("../input/Kannada-MNIST/test.csv")

#get the label
Y_train = mnist_train["label"]

X_train = mnist_train.drop(labels = ["label"], axis=1)

mnist_test = mnist_test.drop(labels = ["id"], axis=1)


#free some space used earlier
del mnist_train
# del mnist_val

#check for null values if any
X_train.isnull().any().describe()
mnist_test.isnull().any().describe()


#Normalize the data
X_train = X_train/255.0
mnist_test = mnist_test/255.0

# Reshape image in 3 dimensions (height = 28px, width = 28px , canal = 1)
X_train = X_train.values.reshape(-1,28,28,1)

# Encode labels to one hot vectors (ex : 2 -> [0,0,1,0,0,0,0,0,0,0])
Y_train = to_categorical(Y_train, num_classes=10)

X_train, X_val, Y_train, Y_val = train_test_split(X_train, Y_train, test_size=0.1, random_state=2)

# Set the CNN model 

def create_model():

    model = Sequential()

    model.add(Conv2D(filters = 64, kernel_size = (5,5),padding = 'Same', 
                     activation ='relu', input_shape = (28,28,1)))
    model.add(Conv2D(filters = 64, kernel_size = (5,5),padding = 'Same', 
                     activation ='relu'))
    model.add(MaxPool2D(pool_size=(2,2)))
    model.add(Dropout(0.4))

    model.add(Conv2D(filters = 64, kernel_size = (5,5),padding = 'Same', 
                     activation ='relu'))
    model.add(Conv2D(filters = 64, kernel_size = (5,5),padding = 'Same', 
                     activation ='relu'))
    model.add(MaxPool2D(pool_size=(2,2)))
    model.add(Dropout(0.4))
    model.add(Flatten())
    model.add(Dense(256, activation = "relu"))
    model.add(Dropout(0.5))
    model.add(Dense(128, activation = "relu"))
    model.add(Dropout(0.4))
    model.add(Dense(10, activation = "softmax"))

    model.summary()

    # Define the optimizer
    optimizer = RMSprop(lr=0.0001, rho=0.9, epsilon=1e-08, decay=0.0)

    # Compile the model
    model.compile(optimizer = optimizer , loss = "categorical_crossentropy", metrics=["accuracy"])

    return model

# Set a learning rate annealer
learning_rate_reduction = ReduceLROnPlateau(monitor='val_loss', 
                                            patience=3, 
                                            verbose=1, 
                                            factor=0.40, 
                                            min_lr=0.000001)

# With data augmentation to prevent overfitting (accuracy 0.99286)

datagen = ImageDataGenerator(
        featurewise_center=False,  # set input mean to 0 over the dataset
        samplewise_center=False,  # set each sample mean to 0
        featurewise_std_normalization=False,  # divide inputs by std of the dataset
        samplewise_std_normalization=False,  # divide each input by its std
        zca_whitening=False,  # apply ZCA whitening
        rotation_range=10,  # randomly rotate images in the range (degrees, 0 to 180)
        zoom_range = 0.1, # Randomly zoom image 
        width_shift_range=0.1,  # randomly shift images horizontally (fraction of total width)
        height_shift_range=0.1,  # randomly shift images vertically (fraction of total height)
        horizontal_flip=False,  # randomly flip images
        vertical_flip=False)  # randomly flip images


#create the model
model = create_model()
datagen.fit(X_train)
epochs = 30 # 
batch_size = 256

# Fit the model
history = model.fit_generator(datagen.flow(X_train,Y_train, batch_size=batch_size),
                              epochs = epochs, validation_data = (X_val,Y_val),
                              verbose = 2, steps_per_epoch=X_train.shape[0] // batch_size
                              , callbacks=[learning_rate_reduction])


# predict results
results = model.predict(mnist_test.values.reshape(-1,28,28,1)) 

# select the indix with the maximum probability
results = np.argmax(results,axis = 1)

results = pd.Series(results,name="label")

submission = pd.concat([pd.Series(range(0,mnist_test.shape[0]),name = "id"),results],axis = 1)

submission.to_csv("submission.csv",index=False)