machine_learn/integr/my_ada_boost.py

# -*- encoding:utf-8 -*-
from sklearn.datasets import load_wine, load_iris, load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
import numpy as np
from tree.my_tree import MyDT


def read_data():
    wine = load_breast_cancer()
    Xtrain, Xtest, Ytrain, Ytest = train_test_split(wine.data,wine.target,test_size=0.3)
    for i in range(len(Ytrain)):
        if Ytrain[i] == 0:
            Ytrain[i] = -1
    for i in range(len(Ytest)):
        if Ytest[i] == 0:
            Ytest[i] = -1
    return Xtrain, Xtest, Ytrain, Ytest


def fit(Xtrain, Ytrain):
    # 设置数据初始权重
    w = np.ones(len(Ytrain))/Ytrain.shape[0]
    # 决策树权重
    alphas = []
    # 决策树数组
    trees = []
    pn = 1/len(Ytrain)
    # print(Ytest)

    for i in range(30):
        # 训练决策树
        clf = MyDT(criterion="gini", max_features=1, max_depth=1,
                                     )  # 实例化，criterion不写的话默认是基尼系数

        clf.fit(Xtrain, Ytrain, w)
        Ypredit = clf.predict(Xtrain)

        error = np.dot(Ypredit != Ytrain, w)
        p_error = sum(Ypredit != Ytrain)/Ytrain.shape[0]

        if error > 0.5:
            continue
        if error == 0:
            error = 0.001

        print("第", i, "轮错误率", p_error, error)
        alpha = 0.5*np.log2(1/error - 1)

        # 更新权重
        for j in range(Ytrain.shape[0]):
            w[j] = w[j]*np.exp(-alpha*Ytrain[j]*Ypredit[j])
        sum_w = sum(w)
        w = w/sum_w

        alphas.append(alpha)
        trees.append(clf)
    return trees, alphas


def cmp_predict(trees, alphas, Xtrain, Ytrain):
    predicts = []
    for tree in trees:
        predicts.append(tree.predict(Xtrain))

    # 结果加权
    result = np.zeros(len(Xtrain), float)
    for p in predicts:
        r = 0
        for w_alpha in alphas:
            r += w_alpha * p
        result = result + r

    # print("sign前:" , result)
    result = np.sign(result)
    # print("sign后:", result)

    # print(1- sum(np.bitwise_xor(Ytest, result))/len(result))
    # print(result == Ytest)
    x = len([i for i in result == Ytrain if i])/len(result)
    x = x + 0.5*len([i for i in result if i==0])/len(result)
    print(x)

    w = np.array([1/len(Xtrain) for i in range(len(Xtrain))])
    Ypredit = trees[0].predict(Xtrain)
    p_error = sum(Ypredit != Ytrain)/Ytrain.shape[0]
    print(1-p_error)


if __name__ == '__main__':
    Xtrain, Xtest, Ytrain, Ytest = read_data()
    trees, alphas = fit(Xtrain, Ytrain)

    print(alphas)

    print('训练集上比较')
    cmp_predict(trees, alphas, Xtrain, Ytrain)
    print('测试集上比较')
    cmp_predict(trees, alphas, Xtest, Ytest)