吴恩达机器学习_线性回归模型week2

一、基础知识

1、学习率

2、归一化

利用lab_utils_multi.py里面的函数对数据进行归一化处理，具体详见代码

3、代码

import numpy as np
import matplotlib.pyplot as plt
from lab_utils_multi import (load_house_data,
                    run_gradient_descent,norm_plot,
                    plt_equal_scale,plot_cost_i_w)
from lab_utils_common import dlc            

np.set_printoptions(precision=2)
plt.style.use('deeplearning.mplstyle')

def zscore_norm_features(x):
    """功能: 按列对x进行x-score归一化\n
    返回:\n
    axis的参数用于指定沿着哪个轴进行计算操作\n
    对于一个二维数组（矩阵）来说：axis=0 表示沿着行的方向进行操作;axis=1 表示沿着列的方向进行操作\n
    x_norm: 按列归一化后的x\n
    mu: shape=(n,), 每种特征的均值\n
    sigma: shape=(n,), 每种特征的标准差\n
    """
    mu = np.mean(x,axis=0)
    sigma = np.std(x,axis=0)
    x_norm = (x-mu)/sigma

    return x_norm,mu,sigma





if __name__ == '__main__':
    x_train,y_train = load_house_data()
    x_features = ["size(sqft)","bedrooms","floors","age"]

    # 显示每种特征对房价的影响
    # fig,ax = plt.subplots(1,4,figsize=(12,3),sharey=True)
    # for i in range(len(ax)):
    #     ax[i].scatter(x_train[:,i],y_train)
    #     ax[i].set_xlabel(x_features[i])
    # ax[0].set_ylabel("price (1000's)")
    # plt.title('features vs. price')
    # plt.show()

    # 设置alpha为9.9e-7
    # _,_,hist = run_gradient_descent(x_train,y_train,10,alpha=9.9e-7)
    # plot_cost_i_w(x_train,y_train,hist)

    # alpha设置的小一点: 9e-7
    # _,_,hist = run_gradient_descent(x_train,y_train,10,alpha=9e-7)
    # plot_cost_i_w(x_train,y_train,hist)

    # alpha设置的再小一点: 1e-7
    # _,_,hist = run_gradient_descent(x_train,y_train,10,alpha=1e-7)
    # plot_cost_i_w(x_train,y_train,hist)

    # 查看归一化后的数据分布
    x_norm,mu,sigma = zscore_norm_features(x_train)
    # fig,ax = plt.subplots(1,3,figsize=(12,3))
    # ax[0].scatter(x_train[:,0],x_train[:,3])
    # ax[0].set_xlabel(x_features[0])
    # ax[0].set_ylabel(x_features[3])
    # ax[0].set_title("unormalized")
    # ax[0].axis('equal')

    # x_mean = x_train-mu
    # ax[1].scatter(x_mean[:,0],x_mean[:,3])
    # ax[1].set_xlabel(x_features[0])
    # ax[1].set_ylabel(x_features[3])
    # ax[1].set_title(r'x-$\mu$')
    # ax[1].axis("equal")

    # ax[2].scatter(x_norm[:,0],x_norm[:,3])
    # ax[2].set_xlabel(x_features[0])
    # ax[2].set_ylabel(x_features[3])
    # ax[2].set_title("z-score normalized")
    # ax[2].axis('equal')

    # plt.tight_layout(rect=[0,0.03,1,0.95])
    # fig.suptitle("distribution of features before, during, after normlization")
    # plt.show()

    # print(f'peak to peak range by column in raw x: {np.ptp(x_train,axis=0)}')
    # print(f'peak to peak range by column in normalized x: {np.ptp(x_norm,axis=0)}')

    # 查看peak to peak分布情况
    # fig,ax = plt.subplots(1,4,figsize=(12,3))
    # for i in range(len(ax)):
    #     norm_plot(ax[i],x_train[:,i])
    #     ax[i].set_xlabel(x_features[i])

    # ax[0].set_ylabel("count")
    # fig.suptitle("distribution of features before normalization")
    # # plt.show()

    # fig,ax = plt.subplots(1,4,figsize=(12,3))
    # for i in range(len(ax)):
    #     norm_plot(ax[i],x_norm[:,i])
    #     ax[i].set_xlabel(x_features[i])

    # ax[0].set_ylabel("count")
    # fig.suptitle("distribution of features after normalization")
    # plt.show()

    # 现在使用归一化后的数据，更大的学习率重新做梯度下降优化
    w_norm,b_norm,hist = run_gradient_descent(x_norm,y_train,1000,1.0e-1)
    # 查看预测结果与标签结果
    # m = x_norm.shape[0]
    # yp = np.zeros(m)
    # for i in range(m):
    #     yp[i] = np.dot(x_norm[i],w_norm) + b_norm
    # fig,ax = plt.subplots(1,4,figsize=(12,3),sharey=True)
    # for i in range(len(ax)):
    #     ax[i].scatter(x_train[:,i],y_train,label='target')
    #     ax[i].set_xlabel(x_features[i])
    #     ax[i].scatter(x_train[:,i],yp,color=dlc['dlorange'],label='predict')
    # ax[0].set_ylabel("price")
    # ax[0].legend()
    # fig.suptitle("target versus prediction using z-score normalized model")
    # plt.show()

    # prediction
    # x_house = np.array([1200,3,1,40])
    # x_house_norm = (x_house - mu) - sigma
    # print(x_house_norm)
    # x_house_predict = np.dot(x_house,w_norm) + b_norm
    # print(f'predicted price of a house with 1200 sqrt, \
    #     3 bedrooms, 1 floor 40 years old = ${x_house_predict*1000:.0f}')

    plt_equal_scale(x_train,x_norm,y_train)
    

4. 正规方程

正规方程是一种直接求解线性回归模型参数的数学方法，而不需要像梯度下降那样迭代优化

def normalEqn(X, y):
    theta = np.linalg.inv(X.T @ X) @ X.T @ y  # 计算参数 theta
    return theta

整个计算过程可以理解为对于线性回归的损失函数（均方误差）的最小化，通过直接求解矩阵方程，得到最优的参数 theta，使得损失函数最小化。与梯度下降不同，正规方程是一种解析解，它能够在一次计算中直接得到最优参数，适用于较小的训练数据集和具有正定协方差矩阵的问题。