spb/wp9_uqo Public

1#!/usr/bin/env python32# Author: Simon-Pierre Boucher — contact@spboucher.ai3"""Step 01 — Build the estimation sample.45Reads the raw DuckDB snapshot (``data/raw/realtor_mls_unique.duckdb``),6parses the semi-structured fields, applies the sample restrictions and trims,7and writes ``data/processed/analysis.parquet``.89Usage:  python scripts/01_build_sample.py10"""11import sys12from pathlib import Path1314sys.path.insert(0, str(Path(__file__).resolve().parents[1] / "src"))1516from wp9 import sample  # noqa: E402171819def main() -> None:20    s = sample.build_and_save()21    by_cat = s.groupby("cat").size().to_dict()22    print(f"Estimation sample written: {len(s):,} listings")23    print(f"  by category : {by_cat}")24    print(f"  by province : {s.groupby('prov').size().to_dict()}")25    print(f"  FSA levels  : {s['fsa_c'].nunique():,} (incl. pooled residual categories)")262728if __name__ == "__main__":29    main()30