"""
Competition baseline kit - tabular (CPU only, free).
Handles NaN, categoricals, mixed types. Usage:
    python baseline_tabular.py --train train.csv --test test.csv --target label --out submissions/predictions.csv
"""
import argparse, pandas as pd
from sklearn.ensemble import HistGradientBoostingClassifier, HistGradientBoostingRegressor
from sklearn.preprocessing import LabelEncoder

p = argparse.ArgumentParser()
p.add_argument("--train", required=True)
p.add_argument("--test", required=True)
p.add_argument("--target", required=True)
p.add_argument("--out", default="submissions/predictions.csv")
args = p.parse_args()

train = pd.read_csv(args.train)
test = pd.read_csv(args.test)
y = train[args.target]
X = train.drop(columns=[args.target])
X_test = test.copy()
assert list(X.columns) == list(X_test.columns), "train/test column mismatch"

for c in X.columns:
    if X[c].dtype == object:
        X[c] = X[c].astype("category").cat.codes
        X_test[c] = X_test[c].astype("category").cat.codes
    X[c] = X[c].astype(float)
    X_test[c] = X_test[c].astype(float)

problem = "class" if y.dtype == object or y.nunique() < 30 else "reg"
if problem == "class":
    le = LabelEncoder()
    y = le.fit_transform(y)
    model = HistGradientBoostingClassifier(max_iter=300)
else:
    model = HistGradientBoostingRegressor(max_iter=300)

model.fit(X, y)
pred = model.predict(X_test)
if problem == "class":
    pred = le.inverse_transform(pred.astype(int).astype(object))

import os
os.makedirs(os.path.dirname(args.out) or ".", exist_ok=True)
pd.DataFrame({"prediction": pred}).to_csv(args.out, index=False)
print(f"baseline done -> {args.out} | problem: {problem}")