吴恩达machine-learning-specialization2022第2周的optional lab: multiple linear regression

1. 在第1周的基础上扩展为多元线性回归

$$f_{\mathbf{w},b}(\mathbf{x})=w_0{x}_0+w_1{x}_1+...+w_{n-1}{x}_{n-1}+b\text{\hspace{1em}(1)}$$

1.1 参数

$$\mathbf{w}=\left(\begin{array}{c}{w}_0\\ w_1\\ \cdots \\ w_{n-1}\end{array}\right)，b是常量参数$$

1.2 目标损失函数

$$J(\mathbf{w},b)=\frac{1}{2m}\sum _{i=0}^{m-1}(f_{\mathbf{w},b}({\mathbf{x}}^{(i)})-y^{(i)}{)}^2\text{\hspace{1em}(3)}$$

2. 代码

同样也是需要实现：

1. 损失计算
2. 梯度计算
3. 梯度更新
   这三个方法

2.1 代码

import numpy as np
import copy
import math
import matplotlib.pyplot as plt
plt.style.use("deeplearning.mplstyle")

np.set_printoptions(precision=2)

def predict(x,w,b):
    p = np.dot(x,w)+b
    return p

def compute_cost(X,y,w,b):
    m = X.shape[0]
    cost = 0.0
    for i in range(m):
        f_wb_i = np.dot(X[i],w)+b
        cost = cost + (f_wb_i-y[i])**2
    cost = cost / (2*m)
    return cost

def compute_gradient(X,y,w,b):
    """功能: 计算梯度\n
    return:\n
    dj_dw: (n,)\n
    dj_db: scalar
    """
    m,n = X.shape
    dj_dw = np.zeros((n,))
    dj_db = 0

    for i in range(m):
        err = (np.dot(X[i],w)+b)-y[i]
        for j in range(n):
            dj_dw[j] = dj_dw[j]+err*X[i,j]
        
        dj_db = dj_db + err
    dj_dw = dj_dw / m
    dj_db = dj_db / m

    return dj_dw,dj_db

def gradient_decent(
    X,y,w_in,b_in,cost_func,gradient_func,
    alpha,num_iters):
    # 历史损失值
    J_history = []
    w = copy.deepcopy(w_in)
    b = b_in

    for i in range(num_iters):
        dj_dw,dj_db = gradient_func(X,y,w,b)

        w = w - alpha*dj_dw
        b = b - alpha*dj_db

        if i<100000:
            J_history.append(cost_func(X,y,w,b))
        
        if i%math.ceil(num_iters/10)==0:
            print(f'Iteration {i:4d}: Cost {J_history[-1]:8.2f}')

    return w,b,J_history




if __name__ == '__main__':
    print(f'multiple variable solution start'.center(60,'='))
    
    X_train = np.array([[2104,5,1,45],
                        [1416,3,2,40],
                        [852,2,1,35]])
    y_train = np.array([460,232,178])
    print(f'x.shape={X_train.shape},y.shape={y_train.shape}')

    # 参数初始化
    b_init = 0
    w_init = np.zeros(shape=(4,))

    # x_vec = X_train[0,:]
    # f_wb = predict(x_vec,w_init,b_init)
    # print(f_wb)

    # cost = compute_cost(X_train,y_train,w_init,b_init)
    # print(cost)

    # tmp_dj_dw,tmp_dj_db = compute_gradient(X_train,y_train,w_init,b_init)
    # print(f'dj_dw={tmp_dj_dw}')
    # print(f'dj_db={tmp_dj_db}')

    # hyper params
    iterations = 1000
    alpha = 5e-7
    w_final,b_final,J_hist = gradient_decent(
        X_train,y_train,w_init,b_init,compute_cost,
        compute_gradient,alpha,iterations
    )
    print(f'b,w found by gradient decent: {b_final:.2f}, {w_final}')
    m,_ = X_train.shape
    for i in range(m):
        print(f'prediction: {predict(X_train[i],w_final,b_final):.2f}',
        f'target value={y_train[i]}')


    fig,(ax1,ax2) = plt.subplots(1,2,constrained_layout=True,figsize=(12,4))
    ax1.plot(J_hist)
    ax2.plot(100+np.arange(len(J_hist[100:])),J_hist[100:])
    ax1.set_title("Cost vs. iteration")
    ax2.set_title("Cost vs. iteration(tail)")
    ax1.set_ylabel('Cost')
    ax2.set_ylabel("Cost")
    ax1.set_xlabel("iteration step")
    ax2.set_xlabel('iteration step')
    plt.show()

2.2 结果

==============multiple variable solution start==============
x.shape=(3, 4),y.shape=(3,)
Iteration    0: Cost  2529.46
Iteration  100: Cost   695.99
Iteration  200: Cost   694.92
Iteration  300: Cost   693.86
Iteration  400: Cost   692.81
Iteration  500: Cost   691.77
Iteration  600: Cost   690.73
Iteration  700: Cost   689.71
Iteration  800: Cost   688.70
Iteration  900: Cost   687.69
b,w found by gradient decent: -0.00, [ 0.2   0.   -0.01 -0.07]
prediction: 426.19 target value=460
prediction: 286.17 target value=232
prediction: 171.47 target value=178