Run PCA on Digits

PCA-Digit.py

'''
Full code on: https://github.com/lettergram/PCA
read() obtains every image from training data 
http://abel.ee.ucla.edu/cvxopt/_downloads/mnist.py
img = single image from read()
img[0] - id
img[1] - image of digit
'''

# Normalize the image
m = np.subtract(img[1], np.mean(img[1]))

# Turn 100 samples of every digit 
# (28 by 28 image) and vectorizes it 
matrix[img[0]] = np.reshape(m, -1).T

# Generate a new 784 x 1000 matrix 
# for each centered, vectorized digit
matrix[img[0]] = np.vstack((matrix[img[0]], centered))

# Run PCA on this 784 x 1000 matrix
pca = (mlab.PCA(num.T, standardize=False)).Y.T

# Grab the top attributes/components
pca = pca[0:attrCount]