Files
threshing-atlas/tests/parity/compare_routes.py
T
432e04c2c7 Refactor: lay the repo out by section (core, collection, analysis, planner, site) (#8)
Moves only; no behaviour changes. Every file now lives under the part of the project it
belongs to, each section has a README, and docs/PRIVACY.md collects the rules all parts follow.

  core/        engine.js, graph.js, solver.js, python/clean.py
  collection/  sources/ (form, register, PDF check, daily update), bot/ (importer)
  analysis/    stats/ (Python statistics), network/ (build-graph.js), reports/ (POLICY.md)
  planner/     build-policy.js, policy-report.js
  site/        atlas/ (the page), build.js, qa/
  tests/       core/ (unit), parity/ (JS vs Python), fixtures/

Checked against the previous layout: tests pass; coded runs, test outputs, graph.json and
policy.json are byte-identical; the built site differs only in comments that name file paths.
The published file names Ken's bot downloads (solver.js, graph.js, data/policy.json) are unchanged.

Also removes the undefined/ folder of QA screenshots committed by mistake (the QA scripts now
default to a temp folder) and test.txt, and drops the unused npm workspaces.


Claude-Session: https://claude.ai/code/session_01UCzTd31GphpMXDhQqWrCMQ

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-03 19:46:33 +00:00

61 lines
3.9 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
"""Recompute route rankings in Python (crude Fisher and stratified Mantel-Haenszel) and compare with engine.js."""
import json, subprocess, sys
sys.path.insert(0, "analysis/stats")
import pandas as pd
from stats_core import odds_ratio, mantel_haenszel, bh
d = pd.DataFrame(json.load(open("tests/out/runs_js.json")))
js = json.loads(subprocess.check_output(["node", "-e", """
const E=require('./core/engine.js'),fs=require('fs');const raw=JSON.parse(fs.readFileSync('data/public/form_responses.json'));
const {runs}=E.codeRows(raw.rows,raw.headers);const o={};for(const t of ['see_Blue','bond_Blue','see_shadow','bond_Green']){o[t]=E.rankRoutes(runs,t,10)}
console.log(JSON.stringify(o));"""]))
MISSING = {"Unknown", "Unsure", "Unsure/unknown", "Multiple", "(not recorded)", "(not asked)", "(not asked/blank)", "Unknown/lost count", "Blank/not asked"}
miss = lambda v: not isinstance(v, str) or v == "" or v in MISSING # noqa: E731
def stage(r):
# a route is only ranked from runs whose every step is known (mirrors rankRoutes in engine.js)
if r.scream.startswith("DIED") or miss(r.scream) or miss(r.resolution): cad = None # noqa
else: cad = ("Help → " + r.resolution) if r.scream == "Investigate & help" else (r.scream + " → " + r.resolution)
hour = {"Tree": "Climb the tree", "Valley": "Valley centre"}.get(r.hour)
fs = None if r.scream == "DIED at first sighting" or miss(r.first_sighting) else r.first_sighting
post_null = not isinstance(r.post_path, str)
post = None if post_null or miss(r.post_path) else r.post_path
vals = [None if miss(r.company) else r.company, None if miss(r.direction) else r.direction, fs, cad, post, hour]
if any(v is None for i, v in enumerate(vals) if not (i == 4 and post_null)): return None
return " › ".join(v if v else "—" for v in vals)
def stratum(r):
if r.source != "Form": return "bulk"
return {"Blue": "blue", "Black": "black", "Blue or Black": "blue-or-black"}.get(r.goal, "other")
d["route"] = d.apply(stage, axis=1); d["stratum"] = d.apply(stratum, axis=1)
d["branch"] = d.branch.where(d.branch.notna(), None)
T = {"see_Blue": (lambda x: x.branch == "Passed the Red: meadow", lambda x: x.saw_blue_meadow),
"bond_Blue": (lambda x: x.hour == "Valley", lambda x: (x.outcome == "Bonded") & (x.outcome_color == "Blue")),
"see_shadow": (lambda x: (x.hour == "Tree") & x.branch.notna(), lambda x: x.saw_shadow),
"bond_Green": (lambda x: x.hour == "Tree", lambda x: (x.outcome == "Bonded") & (x.outcome_color == "Green"))}
fails = 0
for t, (pool, ev) in T.items():
P = d[pool(d)].copy(); P["e"] = ev(P).astype(bool)
K, N = int(P.e.sum()), len(P)
rows = {}
for route, g in P[P.route.notna()].groupby("route"):
if len(g) < 10: continue
k, n = int(g.e.sum()), len(g)
orr = odds_ratio(k, n - k, K - k, N - n - (K - k))
tabs = []
for s, sg in P.groupby("stratum"):
inr = sg.route == route
a = int((inr & sg.e).sum()); nin = int(inr.sum()); c = int((~inr & sg.e).sum()); nout = len(sg) - nin
tabs.append([a, nin - a, c, nout - c])
rows[route] = (k, n, orr, mantel_haenszel(tabs))
jr = {r["route"]: r for r in js[t]["rows"]}
if set(jr) != set(rows): print(t, "route sets differ", set(jr) ^ set(rows)); fails += 1
for route, (k, n, orr, mh) in rows.items():
j = jr.get(route)
if not j: continue
if (k, n) != (j["k"], j["n"]) or abs(orr[0] - j["or"]) > 1e-6 * max(1, orr[0]) or abs(orr[3] - j["fisher_p"]) > 1e-6:
print("crude diff", t, route); fails += 1
if mh is None:
if j["mh_or"] is not None: print("mh none diff", t, route); fails += 1
elif j["mh_or"] is None or abs(mh[0] - j["mh_or"]) > 1e-6 * max(1, mh[0]) or abs(mh[3] - j["mh_p"]) > 1e-5:
print("mh diff", t, route, mh, j["mh_or"], j["mh_p"]); fails += 1
print(f"route rankings for {len(T)} outcomes: {'ALL MATCH' if not fails else str(fails) + ' differences'}")
sys.exit(1 if fails else 0)