经典 ML（一）：回归与分类

监督学习两大类

线性回归（回归）

预测一条直线 y = w₁x₁ + w₂x₂ + ... + b：

from sklearn.linear_model import LinearRegression
from sklearn.datasets import fetch_california_housing
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error, r2_score

data = fetch_california_housing()
X_train, X_test, y_train, y_test = train_test_split(data.data, data.target, random_state=42)

model = LinearRegression()
model.fit(X_train, y_train)

preds = model.predict(X_test)
print(f"MSE: {mean_squared_error(y_test, preds):.3f}")
print(f"R² : {r2_score(y_test, preds):.3f}")    # R² 越接近 1 越好
print("系数:", model.coef_)                      # 各特征的权重

逻辑回归（分类）

虽然叫"回归"，但其实是分类——输出概率值再变成 0/1：

from sklearn.linear_model import LogisticRegression

model = LogisticRegression(max_iter=1000)
model.fit(X_train, y_train)

preds = model.predict(X_test)
probs = model.predict_proba(X_test)[:, 1]        # 属于正类的概率

决策树（分类 / 回归都行）

from sklearn.tree import DecisionTreeClassifier

model = DecisionTreeClassifier(max_depth=5, random_state=42)
model.fit(X_train, y_train)
print(model.score(X_test, y_test))

可视化树：

from sklearn.tree import plot_tree
import matplotlib.pyplot as plt

plt.figure(figsize=(20, 8))
plot_tree(model, feature_names=feature_names, class_names=class_names, filled=True)
plt.show()

随机森林：决策树的"三个臭皮匠"

多个决策树投票，比单棵树准很多：

from sklearn.ensemble import RandomForestClassifier
model = RandomForestClassifier(n_estimators=200, random_state=42)
model.fit(X_train, y_train)

XGBoost / LightGBM：表格数据的天花板

2026 工业级表格 ML 标配——比随机森林还好：

pip install xgboost lightgbm

import xgboost as xgb

model = xgb.XGBClassifier(n_estimators=200, max_depth=5)
model.fit(X_train, y_train)
print(model.score(X_test, y_test))

# 特征重要性
xgb.plot_importance(model)

处理结构化表格数据，XGBoost / LightGBM 经常打败深度学习——别先想 PyTorch。

评估指标

分类

from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, confusion_matrix

accuracy_score(y_test, preds)
precision_score(y_test, preds, average="macro")
recall_score(y_test, preds, average="macro")
f1_score(y_test, preds, average="macro")
confusion_matrix(y_test, preds)

回归

from sklearn.metrics import mean_squared_error, mean_absolute_error, r2_score
mean_squared_error(y_test, preds)
mean_absolute_error(y_test, preds)
r2_score(y_test, preds)

选模型的经验法则

下一篇讲聚类和降维——无监督学习。