[Python] gradient descent machine learning

I'm attending cousera class by Andrew Ng and I saw gradient descent with multiple variable. I'm trying to run gradient descent in python.These are my code in python

import sys
import matplotlib.pyplot as plt

def gradient_descent(y_actual, alpha, iterations, m_current, b_current, x_dataset):
if len(y_actual) != len(x_dataset):
print(f"Dofferent lenght of input which is {len(y_actual)} and {len(x_dataset)}")
sys.exit(1)
else:
pass
m_current = m_current
b_current = b_current
m_derivative = 0
b_derivative = 0
new_m = []
new_b = []
itr = []
for z in range(iterations):
for i in range(len(x_dataset)):
m_derivative += (y_actual - ((m_current * x_dataset)+b_current))*x_dataset
b_derivative += (y_actual - ((m_current * x_dataset) + b_current))
itr.append(z)
m_current = m_current - (alpha*(m_derivative/len(x_dataset)))
b_current = b_current - (alpha * (b_derivative / len(x_dataset)))
new_m.append(m_current)
new_b.append(b_current)
m_derivative = 0
b_derivative = 0
return m_current, b_current, itr, new_m, new_b


m, b, iteration, m_history, b_history = gradient_descent([1, 2, 3, 4, 5, 6, 7, 8, 9, 10], 0.0001, 1000, 0, 0, [5, 8, 1, 4, 9, 3, 9, 8, 3, 5])

plt.subplot(121)
plt.plot(iteration, m_history)
plt.subplot(122)
plt.plot(iteration,b_history)
plt.show()


I plot the iteration and relationship between the iteration and the decrease in b which is the intercept and relationship between the iteration and the decrease in m which is the coefficient. This is the graph that I get:



But the problem is that I saw a lot of graphs about gradient descent in Google like this:



What is the problem of my code that give me this result?

Continue reading...