feat(ui): add interactive forecast intelligence UI (PRP-37) (#305)

[w7-mgfcode]

Summary

PRP-37 — Forecast Intelligence Slice C. Operator-facing UI surface for the
PRP-35 V2 feature contract and the PRP-36 model zoo / per-horizon-bucket
metrics / V-aware ops fields. Backend untouched per PRP-37 — every
visible value reads an existing backend response.

Closes #305 · Parent epic #295 · Slice A #299 · Slice B #302

What lands

New surfaces

Page	Affordance
/visualize/forecast	New Train a new model card — segmented family Tabs → model-type Select → V1/V2 Select → conditional feature-pack toggles + train submission.
/visualize/backtest	RMSE tile (when present), Per-horizon-bucket card (table + chart, metric switcher), Baseline vs feature-aware comparison table. Existing model picker switched to family + filtered model.
/visualize/planner	Scenario method badge (model-driven re-forecast vs heuristic adjustment).
/visualize/batch	Sweep preset Select (5 presets) + multi-model × V1/V2 Sweep matrix picker.
/explorer/run-detail	Feature frame panel — V1/V2 chip + per-group collapsible columns + per-column safety chips. Empty state for pre-PRP-35 runs.
/explorer/run-compare	Champion compatibility badge (comparable-run rule) + Feature frame version row in metrics diff.
/ops	Stale-alias card with reason chips (new V mismatch reason), Model Health table with Prev WAPE + Staleness columns, safer Promote dialog — artifact verify + worse-WAPE ack + V-mismatch ack gates.

New files

• frontend/src/lib/feature-frame-utils.ts + test
• frontend/src/lib/horizon-bucket-utils.ts + test
• frontend/src/components/forecast-intelligence/
  • model-family-tabs.tsx + test
  • model-type-select.tsx (+ model-type-utils.ts) + test
  • feature-frame-select.tsx + test
  • feature-groups-toggle.tsx + test
  • horizon-bucket-table.tsx + test
  • feature-frame-panel.tsx + test
  • champion-compatibility-badge.tsx (+ champion-compatibility-utils.ts) + test
  • promote-confirmation-dialog.tsx + test
  • batch-preset-select.tsx (+ batch-preset-utils.ts) + test
  • batch-matrix-picker.tsx + test
• frontend/src/components/charts/backtest-horizon-buckets-chart.tsx + test

Modified files

• frontend/src/types/api.ts — additive PRP-35 / PRP-36 types
  (FeatureFrameVersion, FeatureGroup, FEATURE_GROUP_VALUES,
  FeatureSafetyClass, TrainRequest, BacktestResponse, FoldResult,
  ModelBacktestResult, StaleReason; ModelRun/FeatureMetadataResponse/
  BatchModelConfig/AliasHealth/ModelHealthEntry extended)
• frontend/src/hooks/use-runs.ts — optional featureFrameVersion param
  (cached in query key only; not forwarded to backend)
• Seven page files: pages/visualize/{forecast,backtest,planner,batch}.tsx,
  pages/explorer/{run-detail,run-compare}.tsx, pages/ops.tsx
• docs/user-guide/dashboard-guide.md — references new affordances on
  every touched page
• docs/user-guide/advanced-forecasting-guide.md — new RAG-indexable
  guide covering families, V1 vs V2, feature packs, horizon buckets,
  champion compatibility, stale aliases, safer Promote, batch presets

Validation

• pnpm tsc --noEmit — clean
• pnpm lint — clean (1 pre-existing warning in data-table.tsx,
  unrelated to PRP-37)
• pnpm test --run — 202 tests pass across 27 files
• Backend regression (forecasting + backtesting + registry + ops,
  not integration) — 518 pass
• git diff app/ alembic/ — empty (backend untouched)
• grep -rn "from 'radix-ui'" src — empty (no barrel imports)

Out of scope / deferred

• V2 backtesting dispatch — deferred to feat(forecast): add feature frame v2 (PRP-35) #299
• Known-future-input pill — dropped per Drift ci: add GitHub Actions workflows for phase-1 readiness #3 in the contract probe
  report (no backend is_known_future field exists on any assumption
  schema)
• Dogfood (Task 26) — not run in this PR; requires the local alembic
  stale-revision reset + at least one V2 SUCCESS run seeded, both noted
  in the HANDOFF carry-forwards. The PR is mergeable on CI green; the
  dogfood is a manual smoke-test the next maintainer can run.
• New agent tool / agent_require_approval — unchanged

Test plan

• CI green on dev-target PR (Lint & Format, Type Check, Test, Migration Check)
• After merge — manual dogfood per PRPs/PRP-37-forecast-intelligence-C-interactive-ui.md#Task-26 (train a V2 run + open run-detail / run-compare / ops / batch in the browser)

[sourcery-ai]

Sorry @w7-mgfcode, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 50edb0f2-6d3b-4a86-ba00-677302228309

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/forecast-interactive-ui

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}