|
|
import numpy as np
|
|
|
import matplotlib.pyplot as plt
|
|
|
|
|
|
|
|
|
def gradient_descent_poly_fit(sampleData, degree, learning_rate=0.001, iterations=50000):
|
|
|
def error_function(y_pred, y): # 定义误差函数error_function()
|
|
|
return y_pred - y
|
|
|
def gradient(degree,error,x_normalized,sampleNum,learning_rate): # 定义梯度函数gradient()
|
|
|
# 对每个参数计算梯度
|
|
|
for i in range(degree + 1):
|
|
|
# 计算损失函数对参数的偏导数(梯度)
|
|
|
gradient = np.dot(error, x_normalized**i) * 2 / sampleNum
|
|
|
# 梯度下降更新参数
|
|
|
theta[i] -= learning_rate * gradient
|
|
|
return theta
|
|
|
|
|
|
|
|
|
x = sampleData[:,0]
|
|
|
y = sampleData[:,1]
|
|
|
# 初始化参数(多项式系数)为0
|
|
|
theta = np.zeros(degree + 1)
|
|
|
sampleNum = len(x) # 样本数量
|
|
|
|
|
|
# 归一化x以改善数值稳定性
|
|
|
x_normalized = (x - x.mean()) / x.std()
|
|
|
# 迭代梯度下降
|
|
|
for _ in range(iterations):
|
|
|
# 通过当前参数计算多项式的值
|
|
|
y_pred = np.polyval(theta[::-1], x_normalized)
|
|
|
|
|
|
# 计算预测值与真实值之间的误差
|
|
|
error = error_function(y_pred, y)
|
|
|
# 计算梯度并更新theta
|
|
|
theta = gradient(degree,error,x_normalized,sampleNum,learning_rate)
|
|
|
|
|
|
return theta, x_normalized
|
|
|
|
|
|
# 设置一个较小的学习率和较多的迭代次数
|
|
|
learning_rate = 0.001
|
|
|
iterations = 50000
|
|
|
degree = 4
|
|
|
|
|
|
x = np.array([0, 100, 200, 300, 400, 500, 600, 700, 800, 900])
|
|
|
y = np.array([10, 20, 10, 50, 80, 130, 210, 340, 550, 890])
|
|
|
sampleData = np.array(list(zip(x, y)))
|
|
|
# 运行梯度下降算法
|
|
|
theta, x_normalized = gradient_descent_poly_fit(sampleData, degree, learning_rate, iterations)
|
|
|
|
|
|
# 用拟合的参数计算多项式的值
|
|
|
x_fit_normalized = np.linspace(x_normalized.min(), x_normalized.max(), 100)
|
|
|
y_fit = np.polyval(theta[::-1], x_fit_normalized)
|
|
|
|
|
|
# 反归一化x_fit
|
|
|
x_fit = x_fit_normalized * x.std() + x.mean()
|
|
|
|
|
|
# 绘制原始数据点和拟合的多项式
|
|
|
plt.scatter(x, y, color='red', label='Sample Data')
|
|
|
plt.plot(x_fit, y_fit, label='Polynomial Fit')
|
|
|
plt.legend()
|
|
|
plt.show()
|
|
|
|
|
|
# 输出拟合参数
|
|
|
print(theta)
|
|
|
# if __name__ == '__main__':
|
|
|
# LimitNum = 1000
|
|
|
# sampleData = random_points(20,0, 1000)
|
|
|
# # 这里为了便于观察使用设计好的数据
|
|
|
# # x = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
|
|
|
# # y = np.array([1, 2, 1, 5, 8, 13, 21, 34, 55, 89])
|
|
|
# # sampleData = np.array(list(zip(x, y))) # 将两个一维数组拼接成二维数组
|
|
|
# m = 3
|
|
|
# theta, covariance_matrix = least_square_method(m, sampleData)
|
|
|
# # print(theta)
|
|
|
# curveData = compute_curveData(0,1000,1,theta,m)
|
|
|
# # print("curveData",curveData)
|
|
|
# draw_dots_and_line(curveData,sampleData,0,1000)
|
|
|
|