tds_2

setup.py

n_data_points = len(inputs)
train = int(n_data_points * 0.8)
validate = (n_data_points - train) // 2
test = n_data_points - train - validate

# train the model over the specified number of epochs
for i in range(epochs):
    
    err = 0
    
    for n in range(train): # iterate over the training set
        
        z = np.array(inputs[n])
        t = int(np.sum(z) >= 0) # evaluates to 1 if True and 0 if False
        
        x = np.append(z, -1) # add bias unit to input vector
        
        
        w += epsilon * gradient_descent(t, w, x) # update weights
        err += loss(t, f(x.dot(w))) # record error
        
        
    ep_err.append(err/train) # average training error per-epoch
    
    for n in range(train, train+validate): # validate
        
        z = np.array(inputs[n])
        t = int(np.sum(z) >= 0) # evaluates to 1 if True and 0 if False
        
        x = np.append(z, -1)
        
        # note: no weights update
        err += loss(t, f(x.dot(w)))
        
        
    ep_err.append(err/validate) # average validation error per-epoch