basic_ml.py

import matplotlib.pyplot as plt

original_training_set = [[1, 1000],[2, 5000],[3, 4500],[5, 7500],[8, 10000],[10, 10000]]
training_set = []

for t in original_training_set:
  t.insert(0, 1) # menambahkan intercept term
  training_set.append(t)

m = len(training_set)
n = len(training_set[0]) - 1
theta = [10] * n
alpha = 0.001
iteration = 1000
error = 0.1

def print_h():
  formula = ''
  for j in range(n):
    formula += str(theta[j]) + ' * x{} + '.format(j)
  return formula[:-2]

def h(x):
  result = 0
  for j in range(n):
    result += theta[j] * x[j]
  return result

def cost_function():
  sums = 0
  for i in range(m):
    sums += pow(h(training_set[i]) - training_set[i][-1], 2)
  return 0.5 * sums

def gradient_descent(theta, j):
  partial_derivative = 0
  for data in training_set:
    partial_derivative += (h(data[:-1]) - data[-1]) * data[j] 
  return theta - (alpha * partial_derivative)

init_error = 0
while abs(init_error - cost_function()) > error:
  init_error = cost_function()
  theta_new = [0] * n
  for j in range(n):
    theta_new[j] = gradient_descent(theta[j], j)
  for j in range(n):
    theta[j] = theta_new[j]

print('H(x) = {}'.format(print_h()))

plt.plot(list(x[1] for x in training_set), list(h(x) for x in training_set), 'r')
plt.plot(list(x[1] for x in training_set), list(y[2] for y in training_set), 'bo')
plt.show()