import numpy as np

def sgn(x):
    return np.where( x > 0 , 1 , -1 )

def theta(s):
    return 1 / ( 1 + np.exp( -s ) )

content_train = open( 'hw3_train.dat', 'r' ).read().strip()
content_test = open( 'hw3_test.dat', 'r' ).read().strip()

x_train = np.array( [ [1] + [ float(x) for x in line.split()[:-1] ] for line in content_train.split('\n') ] )
y_train = np.array( [ int(line.split()[-1]) for line in content_train.split('\n') ] )
x_test = np.array( [ [1] + [ float(x) for x in line.split()[:-1] ] for line in content_test.split('\n') ] )
y_test = np.array( [ int(line.split()[-1]) for line in content_test.split('\n') ] )

N_train = len(y_train)
N_test = len(y_test)
T = 2000
Eta = 0.001

w = np.zeros( len( x_train[0] ) )
for _ in range(T):
    grad = np.sum( ( theta( y_train * np.dot( x_train, w ) ) * -y_train ).reshape((N_train, 1)) * x_train, axis=0 ) / N_train
    w = w - Eta * grad

Eout_avg = np.sum( np.not_equal( y_test, sgn( np.dot( x_test , w ) ) ) ) / N_test
print(f'Eout_avg = {Eout_avg}')