README

This data set contains the following:

Xtrain: 11791 training examples of 4's and 9's, each 15 dimensions in length
ytrain: the corresponding labels of the training examples (0 = 4 digit, 1 = 9 digit)

Xtest: 1991 testing examples of 4's and 9's, each 15 dimensions in length used for prediction
ytest: the corresponding ground truth labels used for evaluating performance (0 = 4 digit, 1 = 9 digit)

Q: A matrix of eigenvectors that is used to map the 15-dimensional feature vectors back into image space. For a 15-dimensional feature vector x, this can be done as follows in pseudocode

	1. Map x2 = Q*x
	2. Reshape x2 into a 28x28 matrix and show it as an image

Because the data has had the mean subtracted off, this image will not look exacly like a handwritten digit, but the contours of the 4 or 9 should be clear.