Source identity
| Field | Recorded value |
|---|---|
| Source type | GitHub source |
| Original record | Janady13/missionmatch / db/migrations/0037_slate_display_order.sql |
| Date | Not specified in source |
| Repository | Janady13/missionmatch |
| Revision | 720d47fae476b0d38477ae4401697d372b1ea2bc |
| Branch | feature/opportunity-to-outcome |
| Path or archive member | db/migrations/0037_slate_display_order.sql |
| Original SHA-256 | ed226dec543c794d14da97285b4369d8b83aa262b2bbee01bcf087cff6734071 |
| Git blob SHA | 114dca16043956d3d6795bda03b44ee08eeb4a34 |
| Reading-copy SHA-256 | ed226dec543c794d14da97285b4369d8b83aa262b2bbee01bcf087cff6734071 |
Every recorded occurrence
These 10 records contain identical original source bytes. Their paths and revisions remain part of the archive.
| Original record | Branch or archive | Revision / identifier | Original location |
|---|---|---|---|
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | feature/opportunity-to-outcome | 720d47fae476b0d38477ae4401697d372b1ea2bc | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | collision-deconfliction | 4ee5a09972608ed89f8377c6665542d5f450f0a6 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | master | 1e9249840a4302ec82e123b9c9cb6f75badea3e4 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | mag-substrate | b82c7100a492dec67dfd1eda6b3f53def859be35 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | time-to-fill-survival | b66e1e505863d4f17741a4b9ede73509e45bd6c6 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | presentation-readiness | eb574acaa29d71ce4d0f852572e87c275c79fdb7 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | vendor-intelligence | fe9796a19c261951d3f2565206c0c1b5ac294465 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | phase2-connectors | e038d4c687aec16f8bd9bc4d907eea14d91f8704 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | feature/aioe-upgrade | 2178e7298d0e5d9af83afa12402fd587f0ba6ed8 | Open original |
| Janady13/missionmatch / db/migrations/0037_slate_display_order.sql | capture-readiness | bae6af7cb2997e819174a9e9cfff5e16067a1975 | Open original |
Recovered source transcript
Download complete source text ↓44 lines · 3.5 KB
-- +goose Up
-- Seed the prior only slate_display_order model, the SECOND MAG capability (the first learned display order
-- re ranker). It rides the SAME coefficient store as the win probability model (migration 0036); no new tables.
-- Its OutputKind is display_order, so by construction it can emit only an ordering, never a per candidate
-- score: the frozen decimal engine (internal/modules/matching v1.4.0) stays the only per candidate scorer, and
-- a human still approves every slate member (never auto rejected). The re ranker only permutes the DISPLAY
-- order of an already scored slate so the candidates a human is most likely to approve surface first.
--
-- The features are the frozen engine's six sub scores (clearance, skills, certifications, customer experience,
-- experience, location), NONE a protected attribute or a proxy. The prior only v1 coefficients are EXACTLY the
-- frozen engine's component weights (0.25, 0.25, 0.15, 0.15, 0.10, 0.10), so the re ranker's utility equals the
-- engine's Total and the display order is byte identical to the engine order the day it ships (zero behavior
-- change). As slate_members.decision (approve/reject) labels accrue, cmd/train-model fits a challenger whose
-- learned weights deviate the order only with evidence, and cmd/eval-model gates promotion on held out ranking
-- quality AND a four fifths no adverse impact check recomputed under the challenger ranking. A display order
-- model needs NO calibration bins (it emits an ordering, not a probability), so none are seeded.
INSERT INTO aedt_learned_models (model_key, version, family, output_kind, feature_spec, champion, notes) VALUES (
'slate_display_order',
1,
'pairwise_logistic',
'display_order',
'{"features": [{"name": "clearance_score", "source": "engine_subscore"}, {"name": "skills_score", "source": "engine_subscore"}, {"name": "certs_score", "source": "engine_subscore"}, {"name": "customer_score", "source": "engine_subscore"}, {"name": "experience_score", "source": "engine_subscore"}, {"name": "location_score", "source": "engine_subscore"}]}',
true,
'Prior only v1. The coefficients are the frozen engine component weights, so the display order is byte identical to the engine order the day it ships. Learned pairwise weights light up as slate_members.decision labels accrue. OutputKind display_order, never a candidate score; the frozen v1.4.0 engine stays the only per candidate scorer and a human approves every member.'
) ON CONFLICT (model_key, version) DO NOTHING;
-- The prior only coefficients ARE the frozen engine component weights, so utility = weighted sum = the engine
-- Total and the order is unchanged. The reserved intercept row is 0.0 (an ordering is invariant to the
-- intercept; it is stored only so the store shape matches the win probability model).
INSERT INTO aedt_model_coefficients (model_key, version, feature_name, coefficient) VALUES
('slate_display_order', 1, 'intercept', 0.0000000000),
('slate_display_order', 1, 'clearance_score', 0.2500000000),
('slate_display_order', 1, 'skills_score', 0.2500000000),
('slate_display_order', 1, 'certs_score', 0.1500000000),
('slate_display_order', 1, 'customer_score', 0.1500000000),
('slate_display_order', 1, 'experience_score', 0.1000000000),
('slate_display_order', 1, 'location_score', 0.1000000000)
ON CONFLICT (model_key, version, feature_name) DO NOTHING;
-- +goose Down
DELETE FROM aedt_model_coefficients WHERE model_key = 'slate_display_order';
DELETE FROM aedt_learned_models WHERE model_key = 'slate_display_order';