Skip to content

[T-071a] ModeSummarizer: ranges, characteristic params, medoid, minimal scenario, auto-label #129

Closed

Description

@CodewithJha

Task T-071a · Milestone M6 — Analysis + risk · Size M · Priority P0 · Epic #6

Implements FR-ANA-002 · PR-06

Branch feat/t-071a-modesummarizer-ranges-characteristic-params-medoid — one task = one branch = one PR (doc 32 §6)

Summary

Implement analysis/modes.py ModeSummarizer.summarize() producing FailureModeSummary per doc 09 §3.3: size, primary class & class mix, [q05,q95] ranges vs space, characteristic flag (< characteristic_coverage), medoid run, minimal reproducing scenario (closest-to-nominal failing member), label "<class> | top-3 conditions".

Context / motivation

Modes are what users read in the Casebook; labels must be human-meaningful ('speed high · light low · noise high').

Scope

In scope

  • Physical-unit formatting for labels (log scale for illuminance).
  • ParamRange value object.
  • Label builder function build_label(summary, space) -> str (pure, separately tested).

Implementation guidance

  • Module roboprobe/analysis/modes.py: class ModeSummarizer: def summarize(self, result: ClusteringResult, runs: Sequence[RunRecord], space: ScenarioSpace) -> list[FailureModeSummary] (doc 05 §7).
  • FailureModeSummary (frozen, analysis/types.py): code (assigned later by matching), size, primary_class, class_mix: Mapping[FailureClass, float], ranges: Mapping[str, ParamRange] (unit + physical q05, q95, space_coverage = (q95-q05)/1), characteristic: frozenset[str] (coverage < analysis.characteristic_coverage), medoid_run_id, minimal_run_id, member_run_ids, label, provisional: bool.
  • Medoid: member minimising the sum of Euclidean distances in feature space (use scipy.spatial.distance.cdist; O(n²) acceptable for clusters ≤ 2 000, else sample 500 members with the analysis seed stream).
  • Minimal reproducing scenario: failing member with the smallest normalised distance to the nominal scenario over characteristic parameters (doc 09 §3.3) — this is what engineers replay first.
  • Label: "<CLASS> | p1 high · p2 low · p3 high" with up to 3 characteristic params ranked by smallest coverage; direction word from range midpoint vs nominal (unit space) — words configurable via a small mapping, not inline literals.
  • Physical formatting via ParameterSpec.denormalize + unit; illuminance displayed on log scale.
  • Noise points (label −1) are reported as a pseudo-mode UNCLUSTERED only in the report, never persisted as a failure_modes row.
  • Standards: respect the layering contracts (doc 16 §2) and conventions in doc 34; no hard-coded tunables — thresholds/defaults come from typed settings (doc 27); no magic numbers (ruff PLR2004); public APIs typed (mypy strict) and documented.

Configuration keys

From the configuration reference (doc 27). Keys marked (new) must be added to the typed settings models and appear in the generated reference (T-003c).

  • analysis.characteristic_coverage

Acceptance criteria

  • Golden synthetic campaign → expected summaries.
  • Label format matches doc example.
  • Characteristic flag honours analysis.characteristic_coverage (env override test).
  • Medoid and minimal-scenario selections verified on a hand-built 5-run cluster.
  • Noise points never produce a persisted mode.
  • Provisional flag propagated from fallback clustering.
  • Tests listed below added and green in CI; coverage of new code meets the gate in doc 18 §3.
  • Docs, config reference and CHANGELOG Unreleased updated where behaviour, keys or interfaces changed.
  • Definition of Done (doc 32 §3) satisfied.

Tests required

  • tests/unit/analysis/test_modes.py.
  • Golden JSON snapshot of summaries for the synthetic multi_mode_6d fixture (tests/fixtures/analysis/).

Dependencies

Blocked by: #127 (T-070b)

Blocks: #133 (T-074a), #137 (T-075a)

Parallelization notes

Files / paths owned by this task — avoid editing other paths; shared files (pyproject.toml entry points, cli/main.py, api/routers/__init__.py, router.tsx, docker/compose.yaml, ci.yml) are append-only:

  • backend/src/roboprobe/analysis/modes.py

Parallel with T-071b.

Definition of Done

Per doc 32 §3 and AGENTS.md §5: lint, format, mypy strict and import-linter green; tests added and passing; docs updated; PR template completed with Closes #<this issue>; Conventional Commit messages referencing T-071a (doc 16 §6); commits authored only by the repository owner identity (AGENTS.md §4); CI green; reviewer checklist (doc 32 §5.1) satisfied.

References

Design docs: doc 09 §3.3 · doc 13 §3.5 · doc 02 §3.5 · doc 05 §7

Requirements: FR-ANA-002 · PR-06 (traceability: doc 28)

Metadata

Metadata

Assignees

No one assigned

    Labels

    area:analysisFailure clustering, SHAP, rules, interactionsmilestone:M6M6 — Analysis + riskpriority:P0Critical path / blocks otherssize:M1-3 daystaskChild task of a milestone epictype:featureNew functionality

    Projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions