Day 6: Multiple Linear Regression: Predicting House Prices (https://www.hackerrank.com/challenges/predicting-house-prices/problem)

linear_regression.py

# Enter your code here. Read input from STDIN. Print output to STDOUT
import numpy as np


def cost(y, y_):
    m = y.shape[1]
    cost = 1/(2*m) * np.sum(np.square(y-y_))
    return cost

dimensions = input()

F, H = map(int, dimensions.split())


X = np.array([[float(x) for x in input().split()] for i in range(H)])

y = X[:, -1].reshape(H, 1) # y => (mx1)
X = X[:, :-1] # remove label y -> X => (mxn)
# print(X, y)

W = np.zeros((F, 1)) # W => (nx1)
b = 0 # b => (1x1)

# print(W, b)

num_iter = 1500
for i in range(num_iter):
    # print('X - W', X.shape, W.shape)
    added = np.matmul(X, W)
    # print('added', added, added.shape)
    y_ = added + b
    # print('y', y.shape)
    alpha = 0.02
    # cost = cost(y, y_)
    diff = y_- y # diff => (mx1)
    difference = np.sum(diff, axis=0)
    # print('diff', diff, diff.shape)
    W = W - alpha * 1/H * np.matmul(X.T, diff)
    b = b - alpha * 1/H * difference

# print('W', W)
# print('b', b)
t = int(input())
test = np.array([[float(x) for x in input().split()] for i in range(t)])
y_ = np.matmul(test, W) + b
for i in y_:
    print(i[0])