Hands-On Machine Learning for Algorithmic Trading

Preparing the data

For illustration, we'll use the Fashion MNIST dataset, a modern drop-in replacement for the classic MNIST handwritten digit dataset popularized by Yann LeCun with LeNet in the 1990s. We also relied on this dataset in Chapter 12, Unsupervised Learning.

Keras makes it easy to access the 60,000 train and 10,000 test grayscale samples with a resolution of 28 x 28 pixels:

from keras.datasets import fashion_mnist(X_train, y_train), (X_test, y_test) = fashion_mnist.load_data()X_train.shape, X_test.shape((60000, 28, 28), (10000, 28, 28))

The data contains clothing items from 10 classes. The following figure plots a sample image for each class:

We reshape the data so that each image is represented by a flat one-dimensional pixel ...