Skip to content

feat: align whole-grain definitions across diet sources#21

Merged
koen-vg merged 1 commit into
Sustainable-Solutions-Lab:mainfrom
koen-vg:feat/whole-grain-alignment
Jul 2, 2026
Merged

feat: align whole-grain definitions across diet sources#21
koen-vg merged 1 commit into
Sustainable-Solutions-Lab:mainfrom
koen-vg:feat/whole-grain-alignment

Conversation

@koen-vg

@koen-vg koen-vg commented Jul 2, 2026

Copy link
Copy Markdown
Member

Unify the whole-grain definition on the food-level taxonomy and the GBD exposure basis:

  • Add a maize-whole food (whole_grains, FBS item 2514, USDA FDC 169697) fed from the maize crop, mirroring the wheat and rice whole/refined pairs, so maize-staple regions have a whole-grain food to carry GBD's whole-grain exposure.
  • Refit diet.fbs.whole_grain_shares jointly against GBD per-country whole-grain exposure (population-weighted bounded least squares, countries whose always-whole coarse-grain base exceeds the GBD target excluded as uninformative): flour-wholemeal 0.07, rice-brown 0.01, maize-whole 0.46.
  • Re-split GDD-IA's total cereal energy by each country's FBS cereal composition instead of GDD's own whole/processed split, which counts decorticated millet and sorghum as processed and starved Sahel staples in GDD mode (NER whole grains 27 -> 249 g/day). Cereal masses are now kcal-derived at model-basis group densities and the kcal-target cereal columns carry the aligned split.
  • Disable piecewise food utility explicitly in the feed, food_waste and food_demand calibration configs; their enforced-baseline validation solves reject it, and base configs like central.yaml enable it.
  • Refresh all three calibration artefact sets (default, gdd-ia, gbd-anchored). Stability L1 costs land at or below their previous centres and all cereal food_demand multipliers stay at 1.0 with zero slack, confirming the realignment introduces no supply/demand mismatch.

Unify the whole-grain definition on the food-level taxonomy and the
GBD exposure basis:

- Add a maize-whole food (whole_grains, FBS item 2514, USDA FDC
  169697) fed from the maize crop, mirroring the wheat and rice
  whole/refined pairs, so maize-staple regions have a whole-grain food
  to carry GBD's whole-grain exposure.
- Refit diet.fbs.whole_grain_shares jointly against GBD per-country
  whole-grain exposure (population-weighted bounded least squares,
  countries whose always-whole coarse-grain base exceeds the GBD
  target excluded as uninformative): flour-wholemeal 0.07, rice-brown
  0.01, maize-whole 0.46.
- Re-split GDD-IA's total cereal energy by each country's FBS cereal
  composition instead of GDD's own whole/processed split, which counts
  decorticated millet and sorghum as processed and starved Sahel
  staples in GDD mode (NER whole grains 27 -> 249 g/day). Cereal
  masses are now kcal-derived at model-basis group densities and the
  kcal-target cereal columns carry the aligned split.
- Disable piecewise food utility explicitly in the feed, food_waste
  and food_demand calibration configs; their enforced-baseline
  validation solves reject it, and base configs like central.yaml
  enable it.
- Refresh all three calibration artefact sets (default, gdd-ia,
  gbd-anchored). Stability L1 costs land at or below their previous
  centres and all cereal food_demand multipliers stay at 1.0 with zero
  slack, confirming the realignment introduces no supply/demand
  mismatch.
@koen-vg koen-vg merged commit 20ee56a into Sustainable-Solutions-Lab:main Jul 2, 2026
4 checks passed
@koen-vg koen-vg deleted the feat/whole-grain-alignment branch July 2, 2026 22:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant