Add MK2 population sanity check harness

scripts/mk2_population_sanity_check.py

@@ -0,0 +1,144 @@
+"""Population-level sanity check harness for Engine MK2 schedules."""
+
+from __future__ import annotations
+
+import argparse
+import random
+import statistics
+import sys
+from collections import defaultdict
+from datetime import date
+from pathlib import Path
+from typing import Callable, Dict, Iterable, List, Sequence, Tuple
+
+ROOT = Path(__file__).resolve().parents[1]
+if str(ROOT) not in sys.path:
+    sys.path.insert(0, str(ROOT))
+
+from archetypes import (
+    create_exhausted_parent,
+    create_night_owl_freelancer,
+    create_office_worker,
+)
+from engines.engine_mk2 import EngineMK2
+from models import PersonProfile
+
+ARCHETYPE_FACTORIES: Dict[str, Callable[[], PersonProfile]] = {
+    "office": create_office_worker,
+    "parent": create_exhausted_parent,
+    "freelancer": create_night_owl_freelancer,
+}
+
+
+def pick_archetype(choice: Sequence[str], rng: random.Random) -> str:
+    if not choice:
+        raise ValueError("At least one archetype must be provided")
+    return rng.choice(choice)
+
+
+def build_profile(archetype: str) -> PersonProfile:
+    try:
+        factory = ARCHETYPE_FACTORIES[archetype]
+    except KeyError as exc:
+        raise ValueError(f"Unknown archetype '{archetype}'") from exc
+    return factory()
+
+
+def summarise_events(events: Iterable[Dict[str, object]]) -> Dict[str, float]:
+    totals: Dict[str, int] = defaultdict(int)
+    for event in events:
+        activity = str(event.get("activity"))
+        minutes = int(event.get("duration_minutes", 0))
+        totals[activity] += minutes
+    return {activity: round(minutes / 60.0, 2) for activity, minutes in totals.items()}
+
+
+def compute_min_mean_max(values: Sequence[float]) -> Tuple[float, float, float]:
+    if not values:
+        return (0.0, 0.0, 0.0)
+    return (min(values), statistics.mean(values), max(values))
+
+
+def bucketize(values: Sequence[float], bucket_edges: Sequence[int]) -> List[Tuple[str, int]]:
+    labels: List[Tuple[str, int]] = []
+    if not values:
+        return labels
+
+    buckets = [0 for _ in range(len(bucket_edges) + 1)]
+    for value in values:
+        placed = False
+        for index, edge in enumerate(bucket_edges):
+            if value < edge:
+                buckets[index] += 1
+                placed = True
+                break
+        if not placed:
+            buckets[-1] += 1
+
+    lower = 0
+    for index, count in enumerate(buckets):
+        if index < len(bucket_edges):
+            upper = bucket_edges[index]
+            label = f"{lower}–{upper}h"
+            lower = upper
+        else:
+            label = f">= {lower}h"
+        labels.append((label, count))
+    return labels
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--samples", type=int, default=100, help="Number of synthetic people to generate")
+    parser.add_argument(
+        "--archetypes",
+        nargs="*",
+        default=list(ARCHETYPE_FACTORIES.keys()),
+        help="Subset of archetypes to sample from",
+    )
+    parser.add_argument(
+        "--seed", type=int, default=42, help="Seed for sampling the population and weekly schedules"
+    )
+    args = parser.parse_args()
+
+    rng = random.Random(args.seed)
+    engine = EngineMK2()
+
+    sleep_totals: List[float] = []
+    work_totals: List[float] = []
+    free_totals: List[float] = []
+
+    for index in range(args.samples):
+        archetype = pick_archetype(args.archetypes, rng)
+        profile = build_profile(archetype)
+        week_seed = rng.randint(0, 10_000_000)
+        result = engine.generate_complete_week(
+            profile=profile,
+            start_date=date(2024, 1, 1),
+            week_seed=week_seed,
+        )
+
+        summary = summarise_events(result["events"])
+        sleep_totals.append(summary.get("sleep", 0.0))
+        work_totals.append(summary.get("work", 0.0))
+        free_totals.append(summary.get("free time", 0.0))
+
+    sleep_stats = compute_min_mean_max(sleep_totals)
+    work_stats = compute_min_mean_max(work_totals)
+    free_stats = compute_min_mean_max(free_totals)
+
+    print(f"Population size: {args.samples}")
+    print("Archetypes:", ", ".join(sorted(set(args.archetypes))))
+    print()
+    print("Weekly sleep hours (min/mean/max): {:.2f} / {:.2f} / {:.2f}".format(*sleep_stats))
+    print("Weekly work hours (min/mean/max): {:.2f} / {:.2f} / {:.2f}".format(*work_stats))
+    print("Weekly free-time hours (min/mean/max): {:.2f} / {:.2f} / {:.2f}".format(*free_stats))
+
+    print()
+    print("Sleep distribution (hours/week):")
+    for label, count in bucketize(sleep_totals, [20, 40, 60]):
+        print(f"  {label:<8} : {count}")
+
+
+if __name__ == "__main__":
+    main()