ex3-nn

import numpy as np
import scipy.io
import matplotlib.pyplot as plt

# load data
data = scipy.io.loadmat('machine-learning-ex3\ex3\ex3data1')
# load weights
weights = scipy.io.loadmat('machine-learning-ex3\ex3\ex3weights')
theta1 = weights['Theta1']
theta2 = weights['Theta2']

print('Data shape:')
print(data['X'].shape)
print(data['y'].shape)

# diplay data
idx = np.random.randint(0, data['X'].shape[0], 100)

def displayData(data, m, n):
   fig, ax = plt.subplots(m, n)

   for i in range(m):
      for j in range(n):
         ax[i,j].imshow(data['X'][idx[m*i + j]].reshape(20,20))

#displayData(data, 10, 10)

def sigmoid(z):
    return 1 / (1 + np.exp(-z))

# One-hot encoder
#from sklearn.preprocessing import OneHotEncoder
#encoder = OneHotEncoder(sparse=False)
#y_onehot = encoder.fit_transform(data['y'])

def forward_propagation(X, theta1, theta2):
   m = X.shape[0]

   a1 = np.insert(X, 0, values=np.ones(m), axis=1)
   z2 = a1.dot(theta1.T)
   a2 = np.insert(sigmoid(z2), 0, values=np.ones(m), axis=1)
   z3 = a2.dot(theta2.T)
   h = sigmoid(z3)
   return h

def predict(h):
   m = h.shape[0]
   n = h.shape[1]
   print(m,n)
   ret = np.zeros((m,n))
   for i in range(m):
      for j in range(n):
         if h[i][j] >= 0.5:
            ret[i][j] = 1
         else:
            ret[i][j] = 0

   return ret

h = forward_propagation(data['X'], theta1, theta2)
y_pred = np.array(np.argmax(h, axis=1) + 1)
correct = [1 if a == b else 0 for (a, b) in zip(y_pred, data['y'])]
accuracy = (sum(map(int, correct)) / float(len(correct)))
print('accuracy = {0}%'.format(accuracy * 100))