What can you rent out in your HDB flat?

"""Prep for the 'What can you rent out in your HDB flat?' demo.
Reads the full 'Renting Out of Flats from Jan 2021' table (data.gov.sg) saved as CSV and writes data.json.
Usage: python3 -I prep_rent.py rent.csv out_dir
"""
import csv, json, os, sys, statistics as st
from collections import defaultdict
from decimal import Decimal, ROUND_HALF_UP

src, out_dir = sys.argv[1], sys.argv[2]
os.makedirs(out_dir, exist_ok=True)

TYPES = ["1-ROOM", "2-ROOM", "3-ROOM", "4-ROOM", "5-ROOM", "EXECUTIVE"]
MIN_N = 30           # a town is ranked for a flat type only with at least this many approvals in the 12 months
LAST12 = ("2025-10", "2026-09")
SAME9_A = ("2022-01", "2022-09")
SAME9_B = ("2026-01", "2026-09")


def r10(x):
    """round half up to the nearest 10"""
    return int((Decimal(str(x)) / 10).quantize(Decimal("1"), rounding=ROUND_HALF_UP) * 10)


def pct(v, q):
    v = sorted(v)
    k = (len(v) - 1) * q
    lo, hi = int(k), min(int(k) + 1, len(v) - 1)
    return v[lo] + (v[hi] - v[lo]) * (k - lo)


rows = []
for r in csv.DictReader(open(src)):
    rows.append((r["rent_approval_date"], r["town"], r["flat_type"], float(r["monthly_rent"])))

total = len(rows)
months = sorted(set(m for m, *_ in rows))
D = {"total_rows": total, "first_month": months[0], "last_month": months[-1], "types": {}}

by_t = defaultdict(list)
for m, town, t, rent in rows:
    by_t[t].append((m, town, rent))

for t in TYPES:
    v = by_t.get(t, [])
    d = {"rows": len(v), "share": len(v) / total}
    all_rents = [x[2] for x in v]
    d["all_median"] = r10(st.median(all_rents)) if v else None
    l12 = [x for x in v if LAST12[0] <= x[0] <= LAST12[1]]
    d["l12_rows"] = len(l12)
    if l12:
        rr = [x[2] for x in l12]
        d["l12_median"] = r10(st.median(rr))
        d["l12_p5"] = r10(pct(rr, .05))
        d["l12_p95"] = r10(pct(rr, .95))
    a = [x[2] for x in v if SAME9_A[0] <= x[0] <= SAME9_A[1]]
    b = [x[2] for x in v if SAME9_B[0] <= x[0] <= SAME9_B[1]]
    d["j9_2022_n"], d["j9_2026_n"] = len(a), len(b)
    d["j9_2022_median"] = r10(st.median(a)) if a else None
    d["j9_2026_median"] = r10(st.median(b)) if b else None
    # by year (2021 flagged for COVID in copy; 2026 = Jan to Sep)
    yr = defaultdict(list)
    for m, town, rent in v:
        yr[m[:4]].append(rent)
    d["by_year"] = {y: {"n": len(x), "median": r10(st.median(x))} for y, x in sorted(yr.items())}
    # towns, latest 12 months
    tw = defaultdict(list)
    for m, town, rent in l12:
        tw[town].append(rent)
    d["towns"] = {town: {"n": len(x), "median": r10(st.median(x)), "ranked": len(x) >= MIN_N} for town, x in sorted(tw.items())}
    D["types"][t] = d

# rows per year overall and per month of last year (for approvals-per-year sentence)
yc = defaultdict(int)
for m, *_ in rows:
    yc[m[:4]] += 1
D["rows_by_year"] = dict(sorted(yc.items()))
D["l12_rows_all"] = sum(1 for m, *_ in rows if LAST12[0] <= m <= LAST12[1])
D["towns_all"] = sorted(set(t for _, t, *_ in rows))
D["min_n"] = MIN_N
D["last12"] = LAST12

json.dump(D, open(os.path.join(out_dir, "data.json"), "w"), indent=1)

print("rows", total, "months", months[0], months[-1], "last-12 rows", D["l12_rows_all"])
for t in TYPES:
    d = D["types"][t]
    ranked = [k for k, v in d["towns"].items() if v["ranked"]]
    unr = [k for k, v in d["towns"].items() if not v["ranked"]]
    print(f"{t}: rows {d['rows']} share {d['share']*100:.1f}% | l12 n={d['l12_rows']} median={d.get('l12_median')} p5-p95={d.get('l12_p5')}-{d.get('l12_p95')} | Jan-Sep22 {d['j9_2022_median']} (n={d['j9_2022_n']}) -> Jan-Sep26 {d['j9_2026_median']} (n={d['j9_2026_n']}) | towns ranked {len(ranked)}, too few {len(unr)} {unr}")
    if ranked:
        lo = min(ranked, key=lambda k: d["towns"][k]["median"])
        hi = max(ranked, key=lambda k: d["towns"][k]["median"])
        print("   cheapest", lo, d["towns"][lo], "| dearest", hi, d["towns"][hi])
print(D["rows_by_year"])