线性回归——Python代码实现

import numpy as np

def computer_error_for_give_point(w, b, points): # 计算出 观测值与计算值 之间的误差， 并累加,最后返回 平均误差 loss = 0 for i in range(len(points)): x = points[i, 0] y = points[i, 1] loss += ((w * x + b ) - y) ** 2
return loss/float(len(points))

下面函数只求导一次更新 w 和 b, 后面可用迭代方法多次更新 w， b

def get_gradient(w_current, b_current, points, LearningRate):
N = len(points) w_gradient = 0 b_gradient = 0 for i in range(N): x = points[i, 0] y = points[i, 1] w_gradient += 2/N * ((w_current*x+b_current)-y)x b_gradient += 2/N * ((w_currentx+b_current)-y) new_w = w_current - LearningRate * w_gradient new_b = b_current - LearningRate * b_gradient return new_w,new_b # 以列表的形式返回

def gradeient_descent_run(w, b, points, learn_rate, iteration): points = np.array(points) LearnRate = learn_rate for i in range(iteration): w, b = get_gradient(w, b, points, LearnRate) return w, b

if name == "main": initialize_w = 0 initialize_b = 0 points = [[10, 10], [9, 9], [8, 8], [7, 7], [6, 6], [5, 5], [4, 4], [3, 3], [2, 2], [1, 1]] w, b = gradeient_descent_run(initialize_w, initialize_b, points, 0.005, 100) print(w) print(b)

来源：oschina

链接：https://my.oschina.net/u/4286839/blog/3408152