### 使用 Python 和 ID3 算法对 Iris 数据集进行分类 为了实现基于 ID3 算法的决策树分类器并可视化决策树结构，可以使用 `sklearn` 提供的功能来加载 Iris 数据集，并通过第三方库（如 `graphviz` 或其他工具）生成可视化的决策树图形。以下是完整的解决方案： #### 加载数据集 Iris 数据集可以通过 `sklearn.datasets` 轻松获取。 ```python from sklearn.datasets import load_iris import pandas as pd # Load the Iris dataset data = load_iris() df = pd.DataFrame(data.data, columns=data.feature_names) df['target'] = data.target print(df.head()) ``` #### 构建 ID3 决策树模型 由于 `scikit-learn` 默认不支持原生的 ID3 算法，因此可以选择安装外部库（如 `decision-tree-id3`），或者手动构建逻辑熵计算部分。 以下是一个简单的自定义实现方法[^1]： ```python import numpy as np from collections import Counter import math class Node: def __init__(self, feature=None, threshold=None, left=None, right=None, *, value=None): self.feature = feature self.threshold = threshold self.left = left self.right = right self.value = value def is_leaf_node(self): return self.value is not None class DecisionTree_ID3: def __init__(self, max_depth=100): self.max_depth = max_depth self.root = None def fit(self, X, y): self.n_classes = len(np.unique(y)) self.n_features = X.shape[1] self.root = self._grow_tree(X, y) def _entropy(self, y): hist = np.bincount(y) ps = hist / len(y) return -np.sum([p * np.log2(p) for p in ps if p > 0]) def _information_gain(self, X_column, thresholds, y): parent_entropy = self._entropy(y) # Compute children entropies (weighted average) left_indices = X_column <= thresholds right_indices = ~left_indices n_left, n_right = np.sum(left_indices), np.sum(~left_indices) e_left, e_right = self._entropy(y[left_indices]), self._entropy(y[right_indices]) child_entropy = (n_left / len(y)) * e_left + (n_right / len(y)) * e_right ig = parent_entropy - child_entropy return ig def _best_split(self, X, y): best_feature_idx, best_threshold = None, None max_information_gain = -1 for i in range(self.n_features): X_column = X[:, i] thresholds = np.percentile(X_column, [10, 20, 30, 40, 50, 60, 70, 80, 90]) # Possible split points for t in thresholds: ig = self._information_gain(X_column, t, y) if ig > max_information_gain: max_information_gain = ig best_feature_idx = i best_threshold = t return best_feature_idx, best_threshold def _grow_tree(self, X, y, depth=0): n_samples, n_features = X.shape n_labels = len(np.unique(y)) # Stopping criteria if depth >= self.max_depth or n_labels == 1 or n_samples < 2: leaf_value = self._most_common_label(y) return Node(value=leaf_value) feat_idx, thresh = self._best_split(X, y) if feat_idx is None: leaf_value = self._most_common_label(y) return Node(value=leaf_value) left_indices, right_indices = X[:, feat_idx] <= thresh, X[:, feat_idx] > thresh left_X, left_y = X[left_indices], y[left_indices] right_X, right_y = X[right_indices], y[right_indices] left_child = self._grow_tree(left_X, left_y, depth + 1) right_child = self._grow_tree(right_X, right_y, depth + 1) return Node(feature=feat_idx, threshold=thresh, left=left_child, right=right_child) def predict(self, X): return np.array([self._traverse_tree(x, self.root) for x in X]) def _traverse_tree(self, x, node): if node.is_leaf_node(): return node.value if x[node.feature] <= node.threshold: return self._traverse_tree(x, node.left) return self._traverse_tree(x, node.right) def _most_common_label(self, y): counter = Counter(y) most_common = counter.most_common(1)[0][0] return most_common ``` #### 可视化决策树 要绘制决策树图，可借助 `matplotlib` 和递归遍历节点的方式完成。 ```python import matplotlib.pyplot as plt def visualize_tree(node, depth=0): if node.is_leaf_node(): print(f"{'| '*depth}Leaf: {node.value}") return print(f"{'| '*depth}{data.feature_names[node.feature]} <= {node.threshold:.2f}:") visualize_tree(node.left, depth + 1) print(f"{'| '*depth}{data.feature_names[node.feature]} > {node.threshold:.2f}:") visualize_tree(node.right, depth + 1) model = DecisionTree_ID3(max_depth=3) X = df[data.feature_names].values y = df['target'].values model.fit(X, y) visualize_tree(model.root) ``` #### 输出 IF-THEN 规则 IF-THEN 规则是通过对决策路径的解析得到的。下面展示如何提取这些规则： ```python rules = [] def extract_rules(node, current_rule=""): if node.is_leaf_node(): rule_str = f"{current_rule} THEN class={node.value}" rules.append(rule_str) return new_rule_left = f"{current_rule} AND {data.feature_names[node.feature]} <= {node.threshold:.2f}" if current_rule else \ f"{data.feature_names[node.feature]} <= {node.threshold:.2f}" extract_rules(node.left, new_rule_left) new_rule_right = f"{current_rule} AND {data.feature_names[node.feature]} > {node.threshold:.2f}" if current_rule else \ f"{data.feature_names[node.feature]} > {node.threshold:.2f}" extract_rules(node.right, new_rule_right) extract_rules(model.root) for r in rules: print(r) ``` --- ### 总结 以上代码实现了基于 ID3 的决策树训练过程、可视化以及 IF-THEN 规则提取功能。该方案适用于小型数据集，能够清晰表达特征之间的关系。