Digits 候选模型公平比较

导入当前步骤需要的数值计算、预处理、模型或评价工具。依赖集中写在代码开头，便于复现。

import numpy as np
from sklearn.datasets import load_digits
from sklearn.dummy import DummyClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import StratifiedKFold, cross_validate, train_test_split
from sklearn.neighbors import KNeighborsClassifier
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier

按固定协议切分训练与测试数据；此处的数据边界决定后续哪些样本可以参与 fit。

X, y = load_digits(return_X_y=True)
train, test = train_test_split(np.arange(len(y)), test_size=0.2, stratify=y, random_state=42)
X_dev, y_dev = X[train], y[train]
folds = list(StratifiedKFold(5, shuffle=True, random_state=42).split(X_dev, y_dev))
models = {
    "dummy": DummyClassifier(strategy="most_frequent"),
    "lr": make_pipeline(StandardScaler(), LogisticRegression(max_iter=3000)),
    "knn": make_pipeline(StandardScaler(), KNeighborsClassifier(n_neighbors=5)),
    "tree": DecisionTreeClassifier(max_depth=8, random_state=42),
    "svm": make_pipeline(StandardScaler(), SVC(C=10.0, kernel="rbf", gamma="scale")),
}
results = {}
for name, model in models.items():
    result = cross_validate(model, X_dev, y_dev, cv=folds,
        scoring="f1_macro", return_train_score=True)
    results[name] = result
    print(f"{name}: train={result['train_score'].mean():.3f}, cv={result['test_score'].mean():.3f}, std={result['test_score'].std(ddof=1):.3f}")
winner = max(results, key=lambda name: results[name]["test_score"].mean())
print("selected_from_development:", winner)
print("untouched_test_rows:", len(test))