Skip to content

Add empirical validation test for Harmonizer metrics - #82

Merged
BruinGrowly merged 1 commit into
mainfrom
claude/review-new-docs-01UDv9fzrkTojttRYasTCPdM
Dec 11, 2025
Merged

BruinGrowly merged 1 commit into
mainfrom
claude/review-new-docs-01UDv9fzrkTojttRYasTCPdM

Conversation

@BruinGrowly

Copy link
Copy Markdown
Owner

Creates scripts/empirical_validation.py that tests whether Harmonizer metrics correlate with real technical debt indicators from git history:

  • Analyzes 51 Python files with git history
  • Correlates LJPW dimensions with fix ratio, churn, and commits
  • Results: 75% of correlations match expected direction (STRONG VALIDATION)

Key findings:

  • Complexity (P) vs Churn: +0.449 correlation (complex files change more)
  • Voltage vs Total Commits: +0.468 correlation (active files have more energy)
  • Abstraction (W) vs Fix Ratio: -0.234 correlation (docs reduce bugs)
  • Structure (J) vs Fix Ratio: -0.347 correlation (structure reduces bugs)

Creates scripts/empirical_validation.py that tests whether Harmonizer
metrics correlate with real technical debt indicators from git history:

- Analyzes 51 Python files with git history
- Correlates LJPW dimensions with fix ratio, churn, and commits
- Results: 75% of correlations match expected direction (STRONG VALIDATION)

Key findings:
- Complexity (P) vs Churn: +0.449 correlation (complex files change more)
- Voltage vs Total Commits: +0.468 correlation (active files have more energy)
- Abstraction (W) vs Fix Ratio: -0.234 correlation (docs reduce bugs)
- Structure (J) vs Fix Ratio: -0.347 correlation (structure reduces bugs)
@BruinGrowly
BruinGrowly merged commit 55efe0d into main Dec 11, 2025
5 of 15 checks passed
@BruinGrowly
BruinGrowly deleted the claude/review-new-docs-01UDv9fzrkTojttRYasTCPdM branch December 11, 2025 04:12
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants