Repository navigation

Round 16: spatial methods validated under ACS sampling error (v0.15.0) - #13

Merged

cafeTechne merged 34 commits into

defaultBranchfrom

round16/spatial-validation

Nov 9, 2026

Merged

Round 16: spatial methods validated under ACS sampling error (v0.15.0)#13
cafeTechne merged 34 commits into
defaultBranchfrom
round16/spatial-validation

Commits on Nov 8, 2026
- docs: Round 16 plan (validating spatial methods under ACS sampling error)
  cafeTechne
  committed
  e37ac57
Commits on Nov 9, 2026

Navigation Menu

Repository navigation

Round 16: spatial methods validated under ACS sampling error (v0.15.0) - #13

Round 16: spatial methods validated under ACS sampling error (v0.15.0)#13
cafeTechne merged 34 commits into
defaultBranchfrom
round16/spatial-validation

Commits on Nov 8, 2026

docs: Round 16 plan (validating spatial methods under ACS sampling error)

Commits on Nov 9, 2026

docs(literature): verified full-text reading notes for Round 16 spatial methods

feat(simulation): LISA validation studies under ACS sampling error (Round 16 B1-B3)

chore: rename to Census Research Engine (github.com/cafeTechne/Census-Research-Engine)

feat(replicate_covariance): sampling covariance between areas from VRE replicates

docs(literature): add full-text notes for LeSage & Pace (2014) and Koley & Bera (SSRN preprint)

feat(spatial): direct/indirect/total effects for the lag model (Round 16 A2)

feat(simulation): Round 16 A spatial regression validation studies (info only)

docs(audits): LISA validation report; correct the LISA caveat's characterization (Round 16 B)

feat(simulation): simulated effect intervals and OLS RS tests in spatial_dgp_recovery

Merge Round 16 B: LISA under ACS sampling error (validation; no default change)

docs(literature): full-text notes for Lorincz-Comi 2021, Jung-Thill-Issel 2019, Rogerson 2025, Lacombe-LeSage 2015

docs(audits): Round 16 A spatial regression validation report; effects caveat

docs(audits): SEM lambda and beta across gamma in Juhl's design (A4)

Merge Round 16 A: spatial regression validation; lag-model effects

feat(replicate_covariance): design near-null replicate directions; leak-free held-out comparison

scripts: acs_error_covariance, live study of ACS sampling-error covariance (PLAN_ROUND_16 C1-C2)

scripts: note that the audit CSV is git-ignored

docs(audits): ACS sampling-error covariance between areas, MI/NY tracts and counties (PLAN_ROUND_16 C1-C2)

Merge Round 16 C: ACS sampling-error covariance from VRE replicates

feat(spatial): correct_outcome_error option for the spatial error model (PLAN_ROUND_16 A1)

docs: correct_outcome_error option in README, CHECK_MAP and the agent protocol

feat(http): polite census.gov downloads shared by every file downloader

Merge Round 16: correct_outcome_error option for the spatial error model

scripts(acs_error_covariance): skip unavailable VRE files with the reason

docs(audits): ACS sampling covariance rerun with 2015-2019 tracts

Merge Round 16: polite census.gov downloads; 2015-2019 tract covariance rerun

fix(spatial): controlled-total caveat states the release-dependent sign (2015-2019 households positive)

fix(http,pums,block_weights): validate manual/cached/downloaded files; pace outside the lock; close retried streams

fix(scripts/acs_error_covariance): a zero replicate denominator invalidates the area for that ratio

fix(spatial): A1 studies use the shipped correction; size limit, SE=0, rho bounds, eigenvalue cache, island multipliers

docs(audits): A1 studies rerun on the shipped correct_outcome_error estimator

Merge Round 16 review fixes

release: v0.15.0 (Round 16); gate report; weekly live cron off the hour