Skip to content

Roadmap

The full development plan lives in PLAN.md (Japanese).

This page provides an English summary.

Vision

pycatdap evolves from a CATDAP Python port to a complete AIC-based EDA and ML error analysis library, differentiating itself with:

  • AIC-driven variable relevance ranking
  • AIC-optimal continuous binning
  • Subset search (CATDAP-02) as the primary mechanism
  • Model-agnostic ML error analysis

Release milestones

The EDA + ML error-analysis arc (Phases A→M) has all shipped. The latest release is v0.12.1.

Version Theme Status
v0.2.0 Core CATDAP-01/02 ✅ released
v0.3.0 Plotly backend + univariate EDA ✅ released
v0.4.0 Bivariate EDA (heatmap, association) ✅ released
v0.5.0 profile() + HTML report ✅ released
v0.6.0 target_analysis() + quality suite + measures ✅ released
v0.7.0 Error labeling utilities + datasets ✅ released
v0.8.0 error_analysis() one-call + datasets ✅ released
v0.9.0 Classification / regression error visualization ✅ released
v0.10.0 Calibration with AIC binning ✅ released
v0.11.0 Slice discovery, cohort comparison, drift ✅ released
v0.12.0 LizyStudio integration enablement (Plotly JSON contracts) ✅ released
v0.12.1 discover_error_slices candidate cap (OOM guard) ✅ released
v0.12.2 Housekeeping: quality gates (#34), DivExplorer interop docs (#32), plan reconciliation 🟢 in progress (unreleased)
v0.13.0 Datasets D5 + interop (pysubgroup AICMeasure, DivExplorer schema) 🟢 on develop, unreleased (#25, #31, #32 done; #47 JNcharacter won't-do — GPL, H-0020)
v0.14.0 R numerical cross-validation + performance benchmarks 🟡 planned (#10, #30, #29)
v1.0.0 API stabilization (plotting.*plot.matplotlib.*) 🟡 planned (#33)

Cross-cutting themes

  • Datasets — shipped: #22, #23, #24; planned (v0.13.0): #25, #47
  • Documentation — ✅ all closed: #26, #27, #28
  • Quality — ✅ quality gates #34; planned: benchmarks #29, R-test CI promotion #30
  • Interop#32 DivExplorer: interop guide shipped, full schema fidelity planned (v0.13.0); #31 pysubgroup planned (v0.13.0)

Meta issue

Track overall progress on Issue #11.