# 导入必要的库
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.feature_selection import SelectKBest, f_classif, RFE
from sklearn.ensemble import RandomForestClassifier

# 1. 加载数据
data = load_breast_cancer()
X = pd.DataFrame(data.data, columns=data.feature_names)
y = data.target
print(f"原始特征数量: {X.shape[1]}")

# 2. 数据预处理：标准化（对某些方法很重要）
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# 3. 划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.3, random_state=42)

# --- 过滤法示例：使用方差和ANOVA F值 ---
print("\n--- 过滤法 (Filter Methods) ---")
# 方法1: 移除低方差特征 (方差为0的特征显然无用)
from sklearn.feature_selection import VarianceThreshold
selector_var = VarianceThreshold(threshold=0.5) # 设置方差阈值
X_train_var = selector_var.fit_transform(X_train)
print(f"移除低方差特征后，剩余特征数量: {X_train_var.shape[1]}")

# 方法2: 基于单变量统计检验 (ANOVA F值) 选择前10个特征
selector_kbest = SelectKBest(score_func=f_classif, k=10)
X_train_kbest = selector_kbest.fit_transform(X_train, y_train)
# 查看被选中的特征
selected_features_kbest = X.columns[selector_kbest.get_support()]
print(f"基于ANOVA F值选择的前10个特征: {list(selected_features_kbest)}")

# --- 包装法示例：递归特征消除 (RFE) ---
print("\n--- 包装法 (Wrapper Methods) ---")
# 使用一个基础模型（如逻辑回归）作为评估器
estimator = LogisticRegression(max_iter=1000, random_state=42)
# 创建RFE对象，要求选择10个特征
selector_rfe = RFE(estimator, n_features_to_select=10, step=1)
selector_rfe.fit(X_train, y_train)
selected_features_rfe = X.columns[selector_rfe.support_]
print(f"RFE选择的特征: {list(selected_features_rfe)}")

# --- 嵌入法示例：基于L1正则化的选择 ---
print("\n--- 嵌入法 (Embedded Methods) ---")
# L1正则化 (Lasso) 会将不重要的特征的系数压为0
lasso = LogisticRegression(penalty='l1', solver='liblinear', C=0.1, random_state=42)
lasso.fit(X_train, y_train)
# 查看系数为0的特征（即被“淘汰”的）
coefficients = lasso.coef_[0]
zero_coeff_features = X.columns[coefficients == 0]
print(f"L1正则化置为0的特征 ({len(zero_coeff_features)}个): {list(zero_coeff_features)}")

# 另一种常见的嵌入法：基于树模型的特征重要性
rf = RandomForestClassifier(n_estimators=100, random_state=42)
rf.fit(X_train, y_train)
# 特征重要性排序
importances = rf.feature_importances_
sorted_idx = np.argsort(importances)[::-1]
print("\n基于随机森林的特征重要性排序（前10）:")
for idx in sorted_idx[:10]:
    print(f"{X.columns[idx]}: {importances[idx]:.4f}")