from sklearn.pipeline import Pipeline
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import OneHotEncoder

numeric_features = ["age", "orders_count"]
categorical_features = ["region"]

preprocess = ColumnTransformer(
    transformers=[
        ("num", StandardScaler(), numeric_features),
        ("cat", OneHotEncoder(handle_unknown="ignore"), categorical_features),
    ]
)

pipe = Pipeline([
    ("prep", preprocess),
    ("clf", LogisticRegression(max_iter=1000)),
])

pipe.fit(X_train, y_train)
pipe.score(X_test, y_test)