### ID3算法在Python中处理连续值属性的实现 ID3算法的核心是通过信息增益选择最佳的属性进行节点划分。然而，原始的ID3算法主要针对离散值属性设计。为了处理连续值属性，需要对数据进行离散化处理，通常采用的方法是将连续值划分为多个区间，并基于这些区间计算信息增益。 以下是一个完整的Python实现，展示如何在ID3算法中处理连续值属性并使用信息增益进行节点划分[^1]。 ```python import numpy as np import pandas as pd from math import log2 class Node: def __init__(self, feature=None, threshold=None, label=None): self.feature = feature # 当前节点的特征 self.threshold = threshold # 连续值的划分阈值 self.label = label # 叶子节点的分类标签 self.children = {} # 子节点字典 def entropy(data): """计算数据集的信息熵""" labels = data.iloc[:, -1].value_counts() total = len(data) ent = -sum((count / total) * log2(count / total) for count in labels) return ent def information_gain(data, feature, threshold=None): """计算信息增益""" total_entropy = entropy(data) if threshold is None: # 离散值属性 values = data[feature].unique() weighted_entropy = sum( (len(data[data[feature] == value]) / len(data)) * entropy(data[data[feature] == value]) for value in values ) else: # 连续值属性 left_data = data[data[feature] <= threshold] right_data = data[data[feature] > threshold] weighted_entropy = ( (len(left_data) / len(data)) * entropy(left_data) + (len(right_data) / len(data)) * entropy(right_data) ) return total_entropy - weighted_entropy def find_best_split(data, features, continuous_features): """寻找最佳的划分属性和阈值""" best_feature = None best_threshold = None max_gain = -float('inf') for feature in features: if feature in continuous_features: # 处理连续值属性 sorted_values = sorted(data[feature].unique()) thresholds = [(sorted_values[i] + sorted_values[i+1]) / 2 for i in range(len(sorted_values)-1)] for threshold in thresholds: gain = information_gain(data, feature, threshold) if gain > max_gain: max_gain = gain best_feature = feature best_threshold = threshold else: # 处理离散值属性 gain = information_gain(data, feature) if gain > max_gain: max_gain = gain best_feature = feature best_threshold = None return best_feature, best_threshold, max_gain def build_tree(data, features, continuous_features): """递归构建决策树""" if len(data.iloc[:, -1].unique()) == 1: # 如果所有样本属于同一类 return Node(label=data.iloc[0, -1]) if not features: # 如果没有可用特征 label = data.iloc[:, -1].value_counts().idxmax() return Node(label=label) best_feature, best_threshold, _ = find_best_split(data, features, continuous_features) tree = Node(feature=best_feature, threshold=best_threshold) if best_threshold is not None: # 连续值属性 left_data = data[data[best_feature] <= best_threshold] right_data = data[data[best_feature] > best_threshold] tree.children['<='] = build_tree(left_data, features, continuous_features) tree.children['>'] = build_tree(right_data, features, continuous_features) else: # 离散值属性 for value in data[best_feature].unique(): subset = data[data[best_feature] == value] if len(subset) == 0: # 防止空子集 tree.children[value] = Node(label=data.iloc[:, -1].value_counts().idxmax()) else: tree.children[value] = build_tree(subset, features - {best_feature}, continuous_features) return tree def predict(tree, sample): """使用决策树预测单个样本的类别""" if tree.label is not None: return tree.label feature_value = sample[tree.feature] if tree.threshold is not None: # 连续值属性 if feature_value <= tree.threshold: return predict(tree.children['<='], sample) else: return predict(tree.children['>'], sample) else: # 离散值属性 return predict(tree.children[sample[tree.feature]], sample) # 示例数据集 data = pd.DataFrame({ 'Outlook': ['Sunny', 'Sunny', 'Overcast', 'Rain', 'Rain', 'Rain', 'Overcast', 'Sunny', 'Sunny', 'Rain', 'Sunny', 'Overcast', 'Overcast', 'Rain'], 'Temperature': [85, 80, 83, 70, 68, 65, 64, 72, 69, 75, 75, 72, 81, 71], 'Humidity': [85, 90, 78, 96, 80, 70, 65, 95, 70, 80, 70, 90, 75, 80], 'Wind': ['Weak', 'Strong', 'Weak', 'Weak', 'Weak', 'Strong', 'Strong', 'Weak', 'Weak', 'Weak', 'Strong', 'Strong', 'Weak', 'Strong'], 'Play': ['No', 'No', 'Yes', 'Yes', 'Yes', 'No', 'Yes', 'No', 'Yes', 'Yes', 'Yes', 'Yes', 'Yes', 'No'] }) features = {'Outlook', 'Temperature', 'Humidity', 'Wind'} continuous_features = {'Temperature', 'Humidity'} # 构建决策树 decision_tree = build_tree(data, features, continuous_features) # 测试预测 sample = {'Outlook': 'Sunny', 'Temperature': 70, 'Humidity': 90, 'Wind': 'Weak'} prediction = predict(decision_tree, sample) print(f"预测结果: {prediction}") ``` 上述代码实现了ID3算法，并扩展了对连续值属性的支持。连续值属性通过找到最佳的划分阈值来实现离散化，从而计算信息增益[^2]。