在简单的线性回归中进行简单的梯度下降，代码+注解

梯度下降原理见下图

代码如下：

import numpy as np


def compute_error_for_line_given_points(w, b, points):
    totalError = 0
    #  points是一个2*数据点个数的二维数组
    for i in range(len(points)):
        x = points[i, 0]  # 第i行的第一个值就是x
        y = points[i, 1]  # 第i行的第二个值就是y
        totalError = totalError + (y - (w * x + b)) ** 2
    return totalError / float(len(points))  # 保证精度不缺失
    


def step_gradient(b_current, w_current, points, learningRate):
    b_gradient = 0
    w_gradient = 0  # 对两个参数进行初始化
    N = float(len(points))
    for i in range(0, len(points)):
        x = points[i, 0]  # 第i行的第一个值就是x
        y = points[i, 1]  # 第i行的第二个值就是y
        b_gradient += -(2 / N) * (y - ((w_current * x) + b_current))
        w_gradient += -(2 / N) * x * (y - ((w_current * x) + b_current))
        # 此处除N的作用是抵消累加的效果，因为实际中不存在一条直线进行梯度下降，更多的是
        # 很多数据点，所以直接将每个点的梯度计算出来，然后直接将现在的wb减去lr*减去的梯度
    new_b = b_current - (learningRate * b_gradient)
    new_w = w_current - (learningRate * b_gradient)
    return [new_b, new_w]


def gradient_descent_runner(points, starting_b, starting_m,
                            learning_rate, num_iterating):
    # num_iterating是迭代次数
    b = starting_b
    m = starting_m
    for i in range(num_iterating):
        b, m = step_gradient(b, m, np.array(points), learning_rate)
    return [b, m]


def run():
    points = np.genfromtxt("data.csv", delimiter=',')
    learning_rate = 0.0001
    initial_b = 0
    initial_m = 0
    num_iterating = 1000
    print(f"Staring gradient descent at b={initial_b}, m={initial_m}, "
          f"error={compute_error_for_line_given_points(initial_b, initial_m, points)}")
    print("Running>>>>>>>>>>>>>>>>>>>>>>")
    [b, m] = gradient_descent_runner(points, initial_b, initial_m, learning_rate, num_iterating)
    print(f"After {num_iterating} iterations b={b}, m={m}, error={compute_error_for_line_given_points(b, m, points)}")


if __name__ == "__main__":
    run()