MultipleLR.py

import numpy as np #type: ignore

# y = 2a + 3b + 4
train_x = np.array([[0, 0], [0, 1], [1, 0], [1, 1], [2, 0], [0, 2], [2, 2], [3, 0], [0, 3], [3, 3]])
train_y = np.array([4, 7, 6, 9, 8, 10, 14, 10, 13, 19])
test = np.array([[4, 0], [0, 4], [4, 4], [5, 0], [0, 5], [5, 5]])
len(train_x)

#Cost Function
def cost_func(w, b):
    total = 0
    for i in range(10):
        total += ((np.dot(w, train_x[i]) + b) - train_y[i])**2
    print(f'Current value: w = {w}, b = {b}')
    print(f'Cost Function: {total/(2*len(train_x))}')
    return total/(2*len(train_x))

#Gradient Descent
a = 0.01
def gradient_descent_b(b, CostDer_b):
    temp_b = b - a*CostDer_b
    return temp_b
def gradient_descent_w(w, CostDer_w):
    X = np.multiply(a, CostDer_w)
    temp_w = np.subtract(w, X)
    return temp_w

def PartDer_w(w, b):
    grads = []
    for j in range(2):
        total = 0
        for i in range(10):
            total += ((np.dot(w, train_x[i]) + b) - train_y[i]) * train_x[i][j]
        grads.append(total / len(train_x))
    return np.array(grads)

def PartDer_b(w, b):
    total = 0
    for i in range(len(train_x)):
        total += ((np.dot(train_x[i], w) + b) - train_y[i])
    return (total/len(train_y))

w = np.array([0, 0])
b = 0
for _ in range(5000):
    temp_w = gradient_descent_w(w, PartDer_w(w, b))
    temp_b = gradient_descent_b(b, PartDer_b(w, b))
    w = temp_w
    b = temp_b
    current_cost = cost_func(w, b)

for _ in range(6):
    print(f"Value for {test[_]}: {np.dot(w, test[_]) + b}")