决策树c4.5 python代码 txt

决策树是一种常用的机器学习算法，可以用来进行分类和回归问题的预测。C4.5是一种决策树算法的改进版，它采用了信息增益比来选择最优的划分属性。

以下是一个用Python实现C4.5算法的代码示例：

import numpy as np
import pandas as pd
from math import log2

# 计算信息熵
def calculate_entropy(data):
    labels = data.iloc[:, -1]
    label_counts = labels.value_counts()
    probs = label_counts / label_counts.sum()
    entropy = (-probs * np.log2(probs)).sum()
    return entropy

# 计算信息增益
def calculate_information_gain(data, feature):
    total_entropy = calculate_entropy(data)
    feature_values = data[feature].unique()
    weighted_entropy = 0
    for value in feature_values:
        subset = data[data[feature]==value]
        subset_entropy = calculate_entropy(subset)
        weighted_entropy += (subset.shape[0] / data.shape[0]) * subset_entropy
    information_gain = total_entropy - weighted_entropy
    return information_gain

# 计算信息增益比
def calculate_information_gain_ratio(data, feature):
    information_gain = calculate_information_gain(data, feature)
    intrinsic_value = 0
    feature_values = data[feature].unique()
    for value in feature_values:
        subset = data[data[feature]==value]
        prob = subset.shape[0] / data.shape[0]
        intrinsic_value += prob * log2(prob)
    information_gain_ratio = information_gain / (-intrinsic_value)
    return information_gain_ratio

# 选择最优的划分属性
def select_best_feature(data):
    features = data.columns[:-1]
    best_feature = None
    best_information_gain_ratio = 0
    for feature in features:
        information_gain_ratio = calculate_information_gain_ratio(data, feature)
        if information_gain_ratio > best_information_gain_ratio:
            best_information_gain_ratio = information_gain_ratio
            best_feature = feature
    return best_feature

# 构建决策树
def build_decision_tree(data):
    labels = data.iloc[:, -1]
    if len(set(labels)) == 1:
        return labels.iloc[0]
    if data.shape[1] == 1:
        return labels.value_counts().idxmax()
    best_feature = select_best_feature(data)
    decision_tree = {best_feature: {}}
    feature_values = data[best_feature].unique()
    for value in feature_values:
        subset = data[data[best_feature]==value].drop(best_feature, axis=1)
        decision_tree[best_feature][value] = build_decision_tree(subset)
    return decision_tree

# 预测新样本
def predict(decision_tree, sample):
    if isinstance(decision_tree, str):
        return decision_tree
    feature = list(decision_tree.keys())[0]
    value = sample[feature]
    sub_tree = decision_tree[feature][value]
    return predict(sub_tree, sample)

# 示例数据
data = pd.DataFrame({
    'Outlook': ['Sunny', 'Sunny', 'Overcast', 'Rainy', 'Rainy', 'Rainy', 'Overcast', 'Sunny', 'Sunny', 'Rainy'],
    'Temperature': ['Hot', 'Hot', 'Hot', 'Mild', 'Cool', 'Cool', 'Cool', 'Mild', 'Cool', 'Mild'],
    'Humidity': ['High', 'High', 'High', 'High', 'Normal', 'Normal', 'Normal', 'High', 'Normal', 'Normal'],
    'Windy': ['False', 'True', 'False', 'False', 'False', 'True', 'True', 'False', 'False', 'False'],
    'Play': ['No', 'No', 'Yes', 'Yes', 'Yes', 'No', 'Yes', 'No', 'Yes', 'Yes']
})

decision_tree = build_decision_tree(data)

# 对新样本进行预测
sample = {'Outlook': 'Sunny', 'Temperature': 'Hot', 'Humidity': 'Normal', 'Windy': 'False'}
prediction = predict(decision_tree, sample)

print("叶子节点的预测结果：", decision_tree)
print("新样本的预测结果：", prediction)

以上代码是一个简单的C4.5算法的实现，通过构建决策树来进行分类预测。代码中使用了Pandas库来处理和分析数据，使用了NumPy库来进行数学运算。首先，定义了计算信息熵、信息增益和信息增益比的函数，然后定义了选择最优划分属性的函数，接着定义了构建决策树和预测新样本的函数。最后，使用一个示例数据来构建决策树，并对一个新样本进行预测。