Repository navigation

Report each pull request's effect on national aggregates on the certified dataset - #2335

Merged

MaxGhenis merged 30 commits into

mainfrom

aggregate-impact-report

Oct 18, 2026

Merged

Report each pull request's effect on national aggregates on the certified dataset#2335
MaxGhenis merged 30 commits into
mainfrom
aggregate-impact-report

Commits on Oct 11, 2026
- Add an aggregate-impact report and pin the certified test dataset
  
  MaxGhenis
  and
  claude
  committed
  1e98071
- Run the aggregate-impact report on every pull request
  
  MaxGhenis
  and
  claude
  committed
  073e808
- Keep the aggregate-impact comment short when nothing changes
  
  MaxGhenis
  and
  claude
  committed
  99909e9
- Document the aggregate-impact report and mandatory test data
  
  MaxGhenis
  and
  claude
  committed
  cbd2fed
- Name the measured refs readably in the aggregate-impact report
  
  MaxGhenis
  and
  claude
  committed
  afa1c2a
- Merge remote-tracking branch 'origin/main' into aggregate-impact-report
  MaxGhenis
  committed
  accf433
- Check that each standard reform moves its target
  
  MaxGhenis
  and
  claude
  committed
  8d3e482
- Keep household-level detail on the runner
  
  MaxGhenis
  and
  claude
  committed
  fd94415
- Close the gaps the independent review found in the report's checks
  
  MaxGhenis
  and
  claude
  committed
  c7a6c96
- Carry each check's failing years in the long-format report
  
  MaxGhenis
  and
  claude
  committed
  bd8bff7
- Stamp a measurement with the tool code that was loaded
  
  MaxGhenis
  and
  claude
  committed
  9b33fe1
Commits on Oct 15, 2026
- Merge main; pin the certified uk-data 1.58.1 and follow main's list changes
  
  MaxGhenis
  and
  claude
  committed
  84f64eb
- Judge counted-once per household, measure flagged heads with others held
  
  MaxGhenis
  and
  claude
  committed
  9505c81
- Calibrate reform responsiveness minimums on the certified pair (d1344)
  
  MaxGhenis
  and
  claude
  committed
  a4b18de
- Suppress small cells and catch gated containment (Sol re-review 2)
  
  MaxGhenis
  and
  claude
  committed
  0b84f8e
- Round totals that contain a suppressed head, so it cannot be recovered
  
  MaxGhenis
  and
  claude
  committed
  d33d407
- Say which model each VAT costing ratio was measured with
  
  MaxGhenis
  and
  claude
  committed
  ea63316
- Protect decile rows and changes as well as levels (Sol re-review 3)
  
  MaxGhenis
  and
  claude
  committed
  b01a7ac
- Withhold exact changes across suppression transitions; weighted diagnostics (Sol re-review 4)
  
  MaxGhenis
  and
  claude
  committed
  17f4e9b
Commits on Oct 16, 2026
- Publish totals without suppressed heads instead of rounding them (Sol re-review 5)
  
  MaxGhenis
  and
  claude
  committed
  7e46336
- Close the channels re-review 6 found around excluded heads (Sol re-review 6)
  
  MaxGhenis
  and
  claude
  committed
  f8dfecd
- Rebuild protected totals from their published parts (Sol re-review 7)
  
  MaxGhenis
  and
  claude
  committed
  0924449
- Close the check channels re-review 8 found around suppressed figures
  
  MaxGhenis
  and
  claude
  committed
  7230ec1
- Render a double count whose coefficient isn't published (Sol delta review 9)
  
  MaxGhenis
  and
  claude
  committed
  228a05a
- Fix the two CI failures; publish each exclusion's sign for W4
  
  MaxGhenis
  and
  claude
  committed
  1191ac5
Commits on Oct 18, 2026

Navigation Menu

Repository navigation

Report each pull request's effect on national aggregates on the certified dataset - #2335

Report each pull request's effect on national aggregates on the certified dataset#2335
MaxGhenis merged 30 commits into
mainfrom
aggregate-impact-report

Commits on Oct 11, 2026

Add an aggregate-impact report and pin the certified test dataset

Run the aggregate-impact report on every pull request

Keep the aggregate-impact comment short when nothing changes

Document the aggregate-impact report and mandatory test data

Name the measured refs readably in the aggregate-impact report

Merge remote-tracking branch 'origin/main' into aggregate-impact-report

Check that each standard reform moves its target

Keep household-level detail on the runner

Close the gaps the independent review found in the report's checks

Carry each check's failing years in the long-format report

Stamp a measurement with the tool code that was loaded

Commits on Oct 15, 2026

Merge main; pin the certified uk-data 1.58.1 and follow main's list changes

Judge counted-once per household, measure flagged heads with others held

Calibrate reform responsiveness minimums on the certified pair (d1344)

Suppress small cells and catch gated containment (Sol re-review 2)

Round totals that contain a suppressed head, so it cannot be recovered

Say which model each VAT costing ratio was measured with

Protect decile rows and changes as well as levels (Sol re-review 3)

Withhold exact changes across suppression transitions; weighted diagnostics (Sol re-review 4)

Commits on Oct 16, 2026

Publish totals without suppressed heads instead of rounding them (Sol re-review 5)

Close the channels re-review 6 found around excluded heads (Sol re-review 6)

Rebuild protected totals from their published parts (Sol re-review 7)

Close the check channels re-review 8 found around suppressed figures

Render a double count whose coefficient isn't published (Sol delta review 9)

Fix the two CI failures; publish each exclusion's sign for W4

Commits on Oct 18, 2026

Merge origin/main into aggregate-impact-report

Record d1601 A ruling for aggregate impact publication

Recalibrate aggregate impact fixtures on model 2.135.6

Merge main 2.135.7 and reconcile fiscal coverage and receipts

Recalibrate certified reform receipts for VAT coverage on main 2.135.7