c4.5算法代码python

C4.5算法是一种经典的决策树算法，下面是一个简单的用Python实现的C4.5算法代码示例：

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

class Node:
    def __init__(self, attribute=None, label=None):
        self.attribute = attribute  # 分裂的属性
        self.label = label  # 叶节点的类别
        self.children = {}  # 子节点

def entropy(data):
    # 计算数据集的熵
    labels = data[:, -1]
    unique_labels, counts = np.unique(labels, return_counts=True)
    probabilities = counts / len(labels)
    entropy = -np.sum(probabilities * np.log2(probabilities))
    return entropy

def split_data(data, attribute_index):
    # 根据属性划分数据集
    unique_values = np.unique(data[:, attribute_index])
    splits = {}
    for value in unique_values:
        splits[value] = data[data[:, attribute_index] == value]
    return splits

def choose_best_attribute(data, attributes):
    # 选择最佳划分属性
    entropy_before_split = entropy(data)
    information_gains = []
    for attribute_index in attributes:
        splits = split_data(data, attribute_index)
        entropy_after_split = 0
        for value, split in splits.items():
            entropy_after_split += (len(split) / len(data)) * entropy(split)
        information_gain = entropy_before_split - entropy_after_split
        information_gains.append(information_gain)
    best_attribute_index = attributes[np.argmax(information_gains)]
    return best_attribute_index

def majority_label(labels):
    # 返回数量最多的类别作为叶节点的类别
    unique_labels, counts = np.unique(labels, return_counts=True)
    majority_label = unique_labels[np.argmax(counts)]
    return majority_label

def create_decision_tree(data, attributes):
    labels = data[:, -1]
    # 如果所有样本属于同一类别，则创建叶节点
    if len(np.unique(labels)) == 1:
        return Node(label=labels[0])
    # 如果没有可用属性了，则创建叶节点，类别为数量最多的类别
    if len(attributes) == 0:
        return Node(label=majority_label(labels))
    best_attribute_index = choose_best_attribute(data, attributes)
    best_attribute = attributes[best_attribute_index]
    decision_tree = Node(attribute=best_attribute)
    splits = split_data(data, best_attribute_index)
    new_attributes = np.delete(attributes, best_attribute_index)
    for value, split in splits.items():
        if len(split) == 0:
            decision_tree.children[value] = Node(label=majority_label(labels))
        else:
            decision_tree.children[value] = create_decision_tree(split, new_attributes)
    return decision_tree

def predict(node, sample):
    if node.label is not None:  # 叶节点
        return node.label
    attribute_value = sample[node.attribute]
    if attribute_value not in node.children:  # 未知取值
        return majority_label(list(node.children.values())[0].labels)
    child_node = node.children[attribute_value]
    return predict(child_node, sample)

# 示例用法
data = np.array([
    [1, 1, 1],
    [1, 1, 0],
    [0, 1, 1],
    [0, 0, 0]
])

attributes = np.array([0, 1])
decision_tree = create_decision_tree(data, attributes)

sample = [1, 0]  # 待预测样本
prediction = predict(decision_tree, sample)
print("预测结果:", prediction)

这是一个简单的实现示例，你可以根据自己的需求进行修改和扩展。希望对你有帮助！