yzhao062 / pyod / 25590628282 / 6
94%
master: 94%

LAST BUILD BRANCH: development
DEFAULT BRANCH: master
Ran
Files 178
Run time 3s
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 93.429% (-0.01%) from 93.439%
25590628282.6

push

github

web-flow 
refactor: decompose ADEngine into private helper modules (#669)

Pure structural refactor. ad_engine.py drops from 1,696 to 1,194 LOC
(30% reduction); 12 private methods + 1 inline consensus block move
into 4 new private modules:

- pyod/utils/_quality_metrics.py - quality math, consensus,
  best detector, feature importance, feature contributions
- pyod/utils/_kb_router.py - KB rule engine + suggest_alternative
- pyod/utils/_detector_factory.py - detector construction from plans
- pyod/utils/_nl_feedback.py - iterate() feedback parsers

Public API byte-identical: 20 public methods, all signatures
unchanged. ADEngine.build_detector preserved as a thin public
wrapper. Helper modules use leading-underscore module names per
PyOD convention; not part of the public API.

Each moved function carries NumPy-style Parameters/Returns docstrings.
4 parallel subagents extracted modules; orchestrator integrated
imports + call sites + thin wrapper sequentially.

Plan-reviewed and implement-reviewed by Codex over 2 rounds; both
Low findings (unused spearmanr import, partial docstring polish)
fixed.

Tests: 122/122 pass (test_ad_engine.py + test_ad_engine_v3.py),
all unmodified.

Spec: docs/superpowers/specs/2026-05-07-adengine-refactor-design.md
Plan: docs/superpowers/plans/2026-05-07-adengine-refactor-plan.md
Parent: PR #668 (fix/adengine-stability-issue-667)

16139 of 17274 relevant lines covered (93.43%)

0.93 hits per line

Source Files on job 25590628282.6