def squared_error_gradient(x_i, y_i, theta):
alpha, beta = theta
return [-2 * error(alpha, beta, x_i, y_i), # alpha partial derivative
-2 * error(alpha, beta, x_i, y_i) * x_i] # beta partial derivative
# choose random value to start
random.seed(0)
theta = [random.random(), random.random()]
alpha, beta = minimize_stochastic(squared_error,
squared_error_gradient,
num_friends_good,
daily_minutes_good,
theta,
0.0001
)
return alpha, beta
def squared_error(x_i, y_i, theta):
alpha, beta = theta
return error(alpha, beta, x_i, y_i)
