Simple implementation of Gradient Descent using Python

simple_gradient_descent.py

import numpy as np
import matplotlib.pyplot as plt


def gradientdescentoptimizer(param, c, learning_rate =0.01):
    # Forward Propagation
    param["cost"] = c[0][0] * param["w"] ** 2 + c[1][0] * param["w"] + c[2][0]
    # Backward Propagation
    param["dw"] = 2 * c[0][0] * param["w"] + c[1][0]
    # Update Omega
    param["w"] = param["w"] - learning_rate * param["dw"]
    return param


if __name__ == '__main__':
    param = {"w": 0., "dw": 0., "cost": 0.}
    learning_rate = 0.01
    cost_data = list()
    # c = np.array([[1.], [-10.], [25.]])
    coefficients = np.array([[1.], [-20.], [100.]])
    for i in range(1000):
        param = gradientdescentoptimizer(param, coefficients, learning_rate)
        cost_data.append(param["cost"])
        print("w after ={} iterations is {}".format(i, param["w"]))
    plt.plot(cost_data)
    plt.xlabel("#iterations")
    plt.ylabel("Cost J")
    plt.title("Gradient Descent")
    plt.show()