西瓜书第三章笔记（待补）

多元线性回归：

逻辑回归（对数几率回归）：

习题：编程实现对率回归，并给出西瓜书数据集3.0α上的结果。
由于数据集前半部分全是正例，后半部分全是负例，直接划分训练集容易全是正例，先调整数据集的顺序。调整后如下图：

数据集80%作为训练集，20%作为测试集。使用梯度下降进行求解:
（书上提到的牛顿法之后补上）

python：

#包初始化
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
# 数据初始化
x = [[0.697,0.460],[0.774,0.376],[0.634,0.264],[0.608,0.318],[0.556,0.215],[0.403,0.237],[0.481,0.149],[0.437,0.211],[0.666,0.091],[0.243,0.267],[0.245,0.057],[0.343,0.099],[0.639,0.161],[0.657,0.198],[0.360,0.370],[0.593,0.042],[0.719,0.103]]
y = [1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0]
n = len(x) // 2
x1 = [x[i // 2] if i % 2 == 0 else x[i // 2 + n] for i in range(len(x))]
y1 = [y[i // 2] if i % 2 == 0 else y[i // 2 + n] for i in range(len(y))]

tt = list()
for i in range(len(x1)):
    tmp = list()
    for j in range(len(x1[i])):
        tmp.append(x1[i][j])
    tmp.append(y1[i])
    tt.append(tmp)
# print(tt)
data_tt = pd.DataFrame(tt,columns=['x1','x2','y'])
print(data_tt)

data_x = pd.DataFrame(x1,columns=['x1','x2'])
data_y = pd.Series(y1)
# print(data_x)
# print(data_y)

# 划分数据
def split_data(data_x,data_y,test_rate = 0.2):
    cnt = int(len(data_y) * (1 - test_rate))
    return data_x[:cnt],data_y[:cnt],data_x[cnt:],data_y[cnt:]
train_x,train_y,test_x,test_y = split_data(data_x,data_y)
# print(f'{train_x}\n{train_y}\n{test_x}\n{test_y}')

# 进行训练
#初始化w和b
data_dim = train_x.shape[1]
data_size = train_x.shape[0]
w = np.zeros((data_dim,))
# print(w)
b = np.zeros((1,))


# help(np.clip)
def _sigmod(z):
    return np.clip(1/(1+np.exp(-z)),1e-8,1-(1e-8))
def _f(x,w,b):
    # print(np.dot(w.T(),x))
    return _sigmod(np.dot(x,w)+b)
def get_gradient(x,y,w,b):
    y_hat = _f(x,w,b)
    e_y = y - y_hat
    w_grad = -np.sum(e_y * x.T,1)
    b_grad = -np.sum(e_y)
    return w_grad , b_grad

def _shuffle(x,y):
    randomize = np.arange(len(x))
    np.random.shuffle(randomize)
    return x[randomize],y[randomize]

def _acc(y_pred,y):
    return 1 - np.mean(np.abs(y_pred - y))

max_iter = 60000
# batch_size = 8
learning_rate = 0.15
step = 1
train_acc = []
test_acc = []

for i in range(max_iter):
    w_grad , b_grad = get_gradient(train_x,train_y,w,b)
    w -= learning_rate / np.sqrt(step) * w_grad
    b -= learning_rate / np.sqrt(step) * b_grad
    step += 1
    train_pred = _f(train_x,w,b)
    test_pred = _f(test_x,w,b)
    train_pred = np.round(train_pred)
    test_pred = np.round(test_pred)
    train_acc.append(_acc(train_pred,train_y))
    test_acc.append(_acc(test_pred,test_y))


print('Train_acc={}'.format(train_acc[-1]))
print('Test_acc={}'.format(test_acc[-1]))
print(w)
# print('w = {}'.format(w))


# Accuracy curve

plt.plot(train_acc)
plt.plot(test_acc)
plt.title('Accuracy')
plt.legend(['train', 'test'])
plt.savefig('acc.png')
plt.show()

LDA：