fashion_mnist.py

import numpy as np
import tensorflow as tf

def fashion_mnist():
    '''This function imports the fashion_mnist data_set from
    keras and filters it into the format I need which is:
     - It splits the data into training and testing splitting
     it into 60,000 and 10,0000 images respectively
     - Each image is 28x28 so it flattens the image out into
     a single vector of dimension 784 (28x28)
     - Since each pixel (i.e. value of each entry in image matrix)
     can range from 0 to 255, the code normalizes it
     - It also does onehotcoding of the training and testing labels'''

    fashion_mnist = tf.keras.datasets.fashion_mnist #importing dataset from keras

    #splitting into its different components
    (train_images, train_labels), (test_images, test_labels) = fashion_mnist.load_data()

    #normalizing and flattening
    train_images = train_images.reshape(-1,28*28)/255.0
    test_images = test_images.reshape(-1,28*28)/255.0

    def onehot(integer_labels):
            #Return matrix whose rows are onehot encodings of integers.
            n_rows = len(integer_labels)
            n_cols = integer_labels.max() + 1
            onehotL = np.zeros((n_rows, n_cols), dtype='uint8')
            onehotL[np.arange(n_rows), integer_labels] = 1
            return onehotL

    train_labels = onehot(train_labels)
    test_labels = onehot(test_labels)

    return train_images, train_labels, test_images, test_labels

if __name__ == '__main__':
    #testing if things work
    train_images, train_labels, test_images, test_labels = fashion_mnist()
    print(train_images[0])