特征选择 vs 特征提取的根本差异：安全攻防中的特征工程策略

from sklearn.datasets import load_breast_cancer
from sklearn.feature_selection import SelectKBest, chi2, mutual_info_classif
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score

# 加载示例数据（乳腺癌数据集，用于演示特征选择）
data = load_breast_cancer()
X, y = data.data, data.target

# 数据划分
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)

# 1. 使用卡方检验进行特征选择
selector_chi2 = SelectKBest(chi2, k=10)
X_train_chi2 = selector_chi2.fit_transform(X_train, y_train)
X_test_chi2 = selector_chi2.transform(X_test)

# 2. 使用互信息进行特征选择
selector_mi = SelectKBest(mutual_info_classif, k=10)
X_train_mi = selector_mi.fit_transform(X_train, y_train)
X_test_mi = selector_mi.transform(X_test)

# 3. 使用随机森林进行特征重要性排序
rf = RandomForestClassifier(n_estimators=100, random_state=42)
rf.fit(X_train, y_train)

# 获取特征重要性
feature_importances = rf.feature_importances_
feature_names = data.feature_names

# 排序特征重要性
sorted_indices = feature_importances.argsort()[::-1]
sorted_features = [feature_names[i] for i in sorted_indices]
sorted_importances = [feature_importances[i] for i in sorted_indices]

# 选择前10个最重要的特征
selected_features_rf = sorted_features[:10]
selected_indices_rf = sorted_indices[:10]
X_train_rf = X_train[:, selected_indices_rf]
X_test_rf = X_test[:, selected_indices_rf]

# 训练模型并评估
rf_chi2 = RandomForestClassifier(n_estimators=100, random_state=42)
rf_chi2.fit(X_train_chi2, y_train)
y_pred_chi2 = rf_chi2.predict(X_test_chi2)
accuracy_chi2 = accuracy_score(y_test, y_pred_chi2)

rf_mi = RandomForestClassifier(n_estimators=100, random_state=42)
rf_mi.fit(X_train_mi, y_train)
y_pred_mi = rf_mi.predict(X_test_mi)
accuracy_mi = accuracy_score(y_test, y_pred_mi)

rf_feature = RandomForestClassifier(n_estimators=100, random_state=42)
rf_feature.fit(X_train_rf, y_train)
y_pred_feature = rf_feature.predict(X_test_rf)
accuracy_feature = accuracy_score(y_test, y_pred_feature)

print(f"卡方检验特征选择后的准确率: {accuracy_chi2:.4f}")
print(f"互信息特征选择后的准确率: {accuracy_mi:.4f}")
print(f"随机森林特征重要性选择后的准确率: {accuracy_feature:.4f}")
print(f"原始特征的准确率: {accuracy_score(y_test, rf.predict(X_test)):.4f}")

# 打印选择的特征
print("\n卡方检验选择的特征索引:", selector_chi2.get_support(indices=True))
print("互信息选择的特征索引:", selector_mi.get_support(indices=True))
print("随机森林选择的特征:", selected_features_rf)

from sklearn.datasets import load_breast_cancer
from sklearn.decomposition import PCA, KernelPCA
from sklearn.manifold import TSNE, UMAP
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.preprocessing import StandardScaler

# 加载示例数据（乳腺癌数据集，用于演示特征提取）
data = load_breast_cancer()
X, y = data.data, data.target

# 数据标准化
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# 数据划分
X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.3, random_state=42)

# 1. 使用PCA进行特征提取
pca = PCA(n_components=10)
X_train_pca = pca.fit_transform(X_train)
X_test_pca = pca.transform(X_test)

# 2. 使用核PCA进行特征提取
kernel_pca = KernelPCA(n_components=10, kernel='rbf', random_state=42)
X_train_kpca = kernel_pca.fit_transform(X_train)
X_test_kpca = kernel_pca.transform(X_test)

# 3. 使用t-SNE进行特征提取（注意：t-SNE通常用于可视化，不用于模型训练）
tsne = TSNE(n_components=2, random_state=42)
X_train_tsne = tsne.fit_transform(X_train)
X_test_tsne = tsne.fit_transform(X_test)  # 注意：t-SNE不支持transform方法，这里为了演示

# 4. 使用UMAP进行特征提取
umap = UMAP(n_components=10, random_state=42)
X_train_umap = umap.fit_transform(X_train)
X_test_umap = umap.transform(X_test)

# 训练模型并评估
rf_pca = RandomForestClassifier(n_estimators=100, random_state=42)
rf_pca.fit(X_train_pca, y_train)
y_pred_pca = rf_pca.predict(X_test_pca)
accuracy_pca = accuracy_score(y_test, y_pred_pca)

rf_kpca = RandomForestClassifier(n_estimators=100, random_state=42)
rf_kpca.fit(X_train_kpca, y_train)
y_pred_kpca = rf_kpca.predict(X_test_kpca)
accuracy_kpca = accuracy_score(y_test, y_pred_kpca)

rf_tsne = RandomForestClassifier(n_estimators=100, random_state=42)
rf_tsne.fit(X_train_tsne, y_train)
y_pred_tsne = rf_tsne.predict(X_test_tsne)
accuracy_tsne = accuracy_score(y_test, y_pred_tsne)

rf_umap = RandomForestClassifier(n_estimators=100, random_state=42)
rf_umap.fit(X_train_umap, y_train)
y_pred_umap = rf_umap.predict(X_test_umap)
accuracy_umap = accuracy_score(y_test, y_pred_umap)

print(f"PCA特征提取后的准确率: {accuracy_pca:.4f}")
print(f"核PCA特征提取后的准确率: {accuracy_kpca:.4f}")
print(f"t-SNE特征提取后的准确率: {accuracy_tsne:.4f}")
print(f"UMAP特征提取后的准确率: {accuracy_umap:.4f}")

# 打印累计方差解释率
print(f"\nPCA累计方差解释率: {sum(pca.explained_variance_ratio_):.4f}")

import pandas as pd
import numpy as np
from sklearn.datasets import make_classification
from sklearn.feature_selection import SelectFromModel
from sklearn.decomposition import PCA
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, f1_score, recall_score
from sklearn.preprocessing import StandardScaler

# 生成高维安全数据集（示例数据）
X, y = make_classification(n_samples=1000, n_features=100, n_informative=20, n_redundant=80, random_state=42)

# 数据标准化
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# 数据划分
X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.3, random_state=42)

# 1. 特征选择：使用随机森林特征重要性
rf = RandomForestClassifier(n_estimators=100, random_state=42)
selector = SelectFromModel(rf, max_features=20)
X_train_selected = selector.fit_transform(X_train, y_train)
X_test_selected = selector.transform(X_test)

# 2. 特征提取：使用PCA
pca = PCA(n_components=20)
X_train_pca = pca.fit_transform(X_train)
X_test_pca = pca.transform(X_test)

# 3. 原始数据（作为对比）
X_train_original = X_train
X_test_original = X_test

# 训练模型并评估
def evaluate_model(X_train, X_test, y_train, y_test, name):
    rf = RandomForestClassifier(n_estimators=100, random_state=42)
    rf.fit(X_train, y_train)
    y_pred = rf.predict(X_test)
    
    accuracy = accuracy_score(y_test, y_pred)
    f1 = f1_score(y_test, y_pred)
    recall = recall_score(y_test, y_pred)
    
    print(f"\n{name} 结果:")
    print(f"准确率: {accuracy:.4f}")
    print(f"F1分数: {f1:.4f}")
    print(f"召回率: {recall:.4f}")
    print(f"特征维度: {X_train.shape[1]}")
    
    return {"name": name, "accuracy": accuracy, "f1": f1, "recall": recall, "dimensions": X_train.shape[1]}

# 评估所有方法
results = []
results.append(evaluate_model(X_train_original, X_test_original, y_train, y_test, "原始数据"))
results.append(evaluate_model(X_train_selected, X_test_selected, y_train, y_test, "特征选择"))
results.append(evaluate_model(X_train_pca, X_test_pca, y_train, y_test, "特征提取（PCA）"))

# 可视化结果
import matplotlib.pyplot as plt

fig, ax = plt.subplots(1, 3, figsize=(15, 5))

metrics = ["accuracy", "f1", "recall"]
titles = ["准确率对比", "F1分数对比", "召回率对比"]

for i, (metric, title) in enumerate(zip(metrics, titles)):
    values = [result[metric] for result in results]
    names = [result["name"] for result in results]
    
    ax[i].bar(names, values, color=['#FF4500', '#32CD32', '#4169E1'])
    ax[i].set_title(title)
    ax[i].set_ylabel(metric)
    ax[i].set_ylim(0.8, 1.0)
    
    # 添加数值标签
    for j, v in enumerate(values):
        ax[i].text(j, v + 0.005, f"{v:.4f}", ha='center', va='bottom')

plt.tight_layout()
plt.show()

# 打印特征维度对比
print("\n特征维度对比:")
for result in results:
    print(f"{result['name']}: {result['dimensions']}维")

# 安装必要的Python库
pip install numpy pandas scikit-learn matplotlib seaborn umap-learn