Codebase Metrics

Step 3 — Coupling & Cohesion Analysis

1. Map import/dependency graphs between modules.
2. Calculate afferent coupling (Ca) — who depends on this module.
3. Calculate efferent coupling (Ce) — what this module depends on.
4. Compute instability index: I = Ce / (Ca + Ce).
5. Assess cohesion by analyzing whether module internals share data and responsibilities.

Step 4 — Health Score Synthesis

1. Combine metrics into a normalized health score per module (0-100).
2. Weight factors: complexity (30%), coupling (25%), size (20%), test coverage (25%).
3. Rank modules from healthiest to most problematic.
4. Identify systemic patterns (e.g., all service layers have high coupling).

Quality Criteria

• Metrics computed from actual code, not estimates [HECHO]
• Thresholds clearly stated and justified
• Outliers specifically identified with file paths
• Trends shown where historical data exists

Anti-Patterns

• Averaging complexity across an entire codebase (hides hotspots)
• Treating all coupling as bad (some coupling is necessary and intentional)
• Measuring only lines of code as a quality proxy
• Ignoring test code in metrics (test quality matters too)