Public data, kept honest · part 4 of 4
insight13 September 2026cost: local CPU; runtime ≈ 2.5 minutes a run
Deliverables: what people asked for
5 runs
The question
Did the change make the dataset more accurate without moving the headline numbers for a reason nobody can explain?
insight
Deliverable (teammate session request): KC Indian-origin population by citizenship from ACS 2019-23 PUMS + SEVIS India share -> data/out/kc_indian_breakdown.json — KC Asian Indian alone 19,104 +-2,079: US-born 6,958 / naturalized 5,438 /…
Runs5
Kept4
Every run
| # | what changed | verdict | KC filings (3 y) | geo resolved % | notes |
|---|---|---|---|---|---|
| 14 | Runtime: fan out state/metro employer tables over 8 forked workers | discarded | 5977 | 99.98 | 706s vs 284s serial - forked pandas workers thrash (copy-on-write page copies / BLAS threads); outputs identical; reverted |
| 18 | Deliverable (user): KC employer table, trailing 36 months, total/new/2026-YTD + consultancy vs direct-employer classification (LCA signals + web research for top 118) ->… | kept | 5912 | 99.537 | no pipeline change; 1535 employers, 5912 LCAs, 2167 new-hire, 637 new in 2026 YTD; consultancies 38% of LCAs |
| 19 | Deliverable (user): nationwide employer table (65k employers, trailing 36 months) + MNC/local consultancy split + combined KC/US app; KC deliverable frozen in… | kept | 5912 | 99.537 | no pipeline change; US 1.64M LCAs, 110,943 employers, 543k new-hire; web-verified 322 employers (KC top 120 + US top 200) |
| 20 | Deliverable (user): F-1 students per KC-metro university (SEVIS CY2024 + IPEDS Fall 2024 + fact books) -> data/out/kc_f1_students.json + third view in the combined app | kept | 5912 | 99.537 | metro best estimate 12,783; UCM Lee's Summit 6,433 SEVIS F-1, UMKC 2,878, Avila 1,191, Park 771; SEVIS publishes only top-500 schools |
| 21 | Deliverable (teammate session request): KC Indian-origin population by citizenship from ACS 2019-23 PUMS + SEVIS India share -> data/out/kc_indian_breakdown.json | kept | 5912 | 99.537 | KC Asian Indian alone 19,104 +-2,079: US-born 6,958 / naturalized 5,438 / non-citizen 6,708; matches ACS table 18,466 +-1,434; MO F-1/M-1 India share 53%, KS… |