梯度下降法实现

最新推荐文章于 2024-10-17 09:49:32 发布

邓冉

最新推荐文章于 2024-10-17 09:49:32 发布

阅读量72

点赞数

文章标签： python

本文链接：https://blog.csdn.net/qq_55321850/article/details/127134919

版权

import numpy as np

theta0 = 0
theta1 = 0
theta2 = 1
alpha = 0.01
X = [[1,2.104,3],[1,1.600,3],[1,2.400,3],[1,1.416,2],[1,3.000,4]]
Y = [4.00,3.30,3.69,2.32,5.40]
def hx(theta0,theta1,theta2,alpha,X,Y):
    j = 0
    der0 = 0
    der1 = 0
    der2 = 0
    for i in range(5):
        hx = theta0 * X[i][0] + theta1 * X[i][1] +theta2 * X[i][2]
        j += 1/2 * (hx - Y[i])**2
        der0 = hx-Y[0]
        der1 = (hx-Y[1]) * X[i][1]
        der2 = (hx-Y[2]) * X[i][2]

        theta0 = theta0 - alpha * der0
        theta1 = theta1 - alpha * der1
        theta2 = theta2 - alpha * der2
    return j,theta0,theta1,theta2

for loop in range(500):
    j,theta0,theta1,theta2 = hx(theta0,theta1,theta2,alpha,X,Y)
    print(j,theta0,theta1,theta2)

实验结果