#!/usr/bin/env python3
# Author: Simon-Pierre Boucher — contact@spboucher.ai
"""Step 02 — The horse race.

Fits the twenty registry models under (i) a random 80/20 holdout and
(ii) a forward-in-time holdout (train < 2025, test 2025–2026), plus the
Box–Cox profile likelihood and summary statistics.

Writes: horserace_random.csv, horserace_temporal.csv, boxcox.csv,
summary_stats.csv, sample_counts.csv

Usage:  python scripts/02_horserace.py
"""
import sys
from pathlib import Path

import pandas as pd

sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "src"))

from wp11 import config, models, sample  # noqa: E402


def main() -> None:
    config.ensure_dirs()
    df = sample.load()
    out = config.REPRODUCED

    # ------------------------------------------------------------ Table 1
    desc = {"price": "Sale price ($)", "area": "Floor area (m2)",
            "lot": "Lot area (m2)", "age": "Building age (years)",
            "floors": "Storeys", "units": "Dwelling units"}
    rows = []
    for var, label in desc.items():
        s = df[var]
        rows.append({"variable": label, "n": len(s), "mean": s.mean(),
                     "sd": s.std(), "p10": s.quantile(.10), "p50": s.median(),
                     "p90": s.quantile(.90)})
    pd.DataFrame(rows).to_csv(out / "summary_stats.csv", index=False)
    counts = {"n_sales": len(df), "n_munis": df["muni"].nunique(),
              "n_grid": df["grid"].nunique(),
              "n_grid5": df["grid5"].nunique()}
    for k, v in df.groupby("prop_class").size().items():
        counts[f"n_{k}"] = int(v)
    pd.Series(counts).to_csv(out / "sample_counts.csv")

    # ------------------------------------------------------------ Box–Cox profile
    d = models.add_derived(df)
    d["sale_year_c"] = d["sale_year"].astype(str)
    train, _ = models.split(d, "random")
    lam, prof = models.fit_boxcox(train, models.NUM_LEVELS,
                                  models.CATS + ["muni", "quarter"])
    prof["chosen"] = prof["lambda"] == lam
    prof.to_csv(out / "boxcox.csv", index=False)
    print(f"Box–Cox lambda* = {lam}")

    # ------------------------------------------------------------ the race
    for scheme in ("random", "temporal"):
        res = models.run_horserace(df, scheme)
        res.to_csv(out / f"horserace_{scheme}.csv", index=False)


if __name__ == "__main__":
    main()