機械学習の全体地図（7）Pythonで書く

from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix, classification_report

# 乳がん診断データ（0=悪性, 1=良性）
X, y = load_breast_cancer(return_X_y=True)

# 最初にテスト用を取り分ける。stratify でクラスの比率を保つ
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, stratify=y, random_state=0)

# 前処理とモデルを1つにまとめる
model = make_pipeline(StandardScaler(), LogisticRegression(max_iter=1000))
model.fit(X_train, y_train)

# テスト用データで1回だけ評価する
pred = model.predict(X_test)
print(confusion_matrix(y_test, pred))  # [[TN, FP], [FN, TP]] の並び
print(classification_report(y_test, pred, digits=3))

from sklearn.datasets import load_diabetes
from sklearn.model_selection import train_test_split
from sklearn.dummy import DummyRegressor
from sklearn.linear_model import Ridge
from sklearn.ensemble import HistGradientBoostingRegressor
from sklearn.metrics import mean_absolute_error, root_mean_squared_error, r2_score

X, y = load_diabetes(return_X_y=True)  # 1年後の病気の進行度（数値）
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)

models = {
    "ベースライン（平均値）": DummyRegressor(),
    "線形（Ridge）": Ridge(),
    "勾配ブースティング": HistGradientBoostingRegressor(max_iter=200, learning_rate=0.05, random_state=0),
}
for name, m in models.items():
    pred = m.fit(X_train, y_train).predict(X_test)
    print(f"{name}: MAE={mean_absolute_error(y_test, pred):.1f} "
          f"RMSE={root_mean_squared_error(y_test, pred):.1f} R2={r2_score(y_test, pred):.3f}")

import numpy as np
import pandas as pd
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import make_pipeline
from sklearn.impute import SimpleImputer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.linear_model import LogisticRegression

df = pd.DataFrame({
    "年齢": [25, 32, np.nan, 51, 46, 29, 38, np.nan],
    "月額": [3000, 5000, 4500, np.nan, 8000, 3500, 6000, 7000],
    "プラン": ["A", "B", "B", "C", "C", "A", None, "B"],
    "解約": [1, 0, 0, 1, 0, 1, 0, 0],
})
X, y = df.drop(columns="解約"), df["解約"]

# 数値列: 中央値で補完＋欠損フラグ → 標準化
num = make_pipeline(SimpleImputer(strategy="median", add_indicator=True), StandardScaler())
# カテゴリ列: 最頻値で補完 → One-hot（未知のカテゴリは無視）
cat = make_pipeline(SimpleImputer(strategy="most_frequent"), OneHotEncoder(handle_unknown="ignore"))
pre = ColumnTransformer([("num", num, ["年齢", "月額"]), ("cat", cat, ["プラン"])])

model = make_pipeline(pre, LogisticRegression()).fit(X, y)
print(pre.get_feature_names_out())

new = pd.DataFrame({"年齢": [40], "月額": [np.nan], "プラン": ["D"]})  # 欠損と未知のカテゴリ
print("解約確率:", model.predict_proba(new)[:, 1].round(3))

import numpy as np
from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import cross_val_score, StratifiedKFold, TimeSeriesSplit
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression

X, y = load_breast_cancer(return_X_y=True)
model = make_pipeline(StandardScaler(), LogisticRegression(max_iter=1000))

# 分類: 層化k分割（各分割でクラス比率をそろえる）
cv = StratifiedKFold(n_splits=5, shuffle=True, random_state=0)
scores = cross_val_score(model, X, y, cv=cv, scoring="f1")
print("F1:", scores.round(3), "平均", scores.mean().round(3))

# 時系列: 常に過去で学び、直後で検証する（シャッフルしない）
for train_idx, test_idx in TimeSeriesSplit(n_splits=3).split(np.arange(12)):
    print("学習", train_idx, "→ 検証", test_idx)

# 同じ顧客の行が複数ある場合は GroupKFold を使い、
# cross_val_score(model, X, y, cv=GroupKFold(5), groups=顧客ID) とする

from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression

X, y = load_breast_cancer(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, stratify=y, random_state=0)

model = make_pipeline(StandardScaler(), LogisticRegression(max_iter=1000))
param = {"logisticregression__C": [0.01, 0.1, 1, 10, 100]}  # 正則化の強さ（小さいほど強い）

search = GridSearchCV(model, param, cv=5, scoring="f1").fit(X_train, y_train)
print("最良の設定:", search.best_params_, "交差検証F1:", round(search.best_score_, 3))
print("テストF1（最後に1回）:", round(search.score(X_test, y_test), 3))

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split, TunedThresholdClassifierCV
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, precision_score, recall_score, average_precision_score

X, y = make_classification(n_samples=3000, weights=[0.95], class_sep=0.5,
                           flip_y=0.02, random_state=0)
X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y, random_state=0)

for cw in [None, "balanced"]:
    m = LogisticRegression(class_weight=cw).fit(X_train, y_train)
    p = m.predict(X_test)
    print(f"class_weight={cw}: 正解率={accuracy_score(y_test, p):.3f} "
          f"適合率={precision_score(y_test, p):.3f} 再現率={recall_score(y_test, p):.3f} "
          f"PR-AUC={average_precision_score(y_test, m.predict_proba(X_test)[:, 1]):.3f}")

# F1が最大になる閾値を交差検証で選ぶ（scikit-learn 1.5以上）
tuned = TunedThresholdClassifierCV(LogisticRegression(), scoring="f1").fit(X_train, y_train)
print("選ばれた閾値:", round(tuned.best_threshold_, 3))

from sklearn.datasets import load_iris
from sklearn.preprocessing import StandardScaler
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
from sklearn.metrics import silhouette_score, adjusted_rand_score

X, y = load_iris(return_X_y=True)  # アヤメ3品種。y はクラスタリングには使わない
Xs = StandardScaler().fit_transform(X)

for k in [2, 3, 4, 5]:
    labels = KMeans(n_clusters=k, n_init=10, random_state=0).fit_predict(Xs)
    print(f"k={k}: シルエット={silhouette_score(Xs, labels):.3f} "
          f"正解との一致ARI={adjusted_rand_score(y, labels):.3f}")

pca = PCA(n_components=2).fit(Xs)
print("2次元に圧縮して残る情報の割合:", pca.explained_variance_ratio_.sum().round(3))

import numpy as np
from sklearn.ensemble import IsolationForest

rng = np.random.default_rng(0)
normal = rng.normal(0, 1, size=(300, 2))    # 正常データ
odd = rng.uniform(4, 6, size=(10, 2))       # 珍しいデータ
X = np.vstack([normal, odd])
is_odd = np.r_[np.zeros(300), np.ones(10)]

iso = IsolationForest(contamination=0.03, random_state=0).fit(X)
flag = iso.predict(X) == -1                  # 異常と判定したもの
print("異常と判定:", flag.sum(), "件")
print("そのうち本当に珍しいデータ:", int((flag & (is_odd == 1)).sum()), "/ 10件")

from sklearn.datasets import load_breast_cancer
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
from sklearn.inspection import permutation_importance

data = load_breast_cancer()
X_train, X_test, y_train, y_test = train_test_split(
    data.data, data.target, test_size=0.2, stratify=data.target, random_state=0)

model = RandomForestClassifier(random_state=0).fit(X_train, y_train)
r = permutation_importance(model, X_test, y_test, n_repeats=10, random_state=0)

for i in r.importances_mean.argsort()[::-1][:5]:
    print(f"{data.feature_names[i]:<25} {r.importances_mean[i]:.4f}")

from sklearn.datasets import make_classification
from sklearn.model_selection import validation_curve
from sklearn.tree import DecisionTreeClassifier

X, y = make_classification(n_samples=500, n_features=20, flip_y=0.15, random_state=0)
depths = [1, 2, 3, 5, 8, 12, None]
train, valid = validation_curve(DecisionTreeClassifier(random_state=0), X, y,
                                param_name="max_depth", param_range=depths, cv=5)
for d, a, b in zip(depths, train.mean(1), valid.mean(1)):
    print(f"深さ={str(d):>4}: 学習={a:.3f} 検証={b:.3f}")