LEM/pkg/lem/analytics.go

package lem

import (
	poindexter "github.com/Snider/Poindexter"
)

// ScoreDistribution wraps Poindexter's DistributionStats for LEM score populations.
type ScoreDistribution = poindexter.DistributionStats

// GrammarAxisStats wraps Poindexter's AxisDistribution for per-feature analysis.
type GrammarAxisStats = poindexter.AxisDistribution

// ComputeScoreDistribution calculates percentile/variance stats over grammar composites.
func ComputeScoreDistribution(scores []GrammarScore) ScoreDistribution {
	vals := make([]float64, len(scores))
	for i, s := range scores {
		vals[i] = s.Composite
	}
	return poindexter.ComputeDistributionStats(vals)
}

// ComputeLEKDistribution calculates percentile/variance stats over LEK scores.
func ComputeLEKDistribution(scores []*HeuristicScores) ScoreDistribution {
	vals := make([]float64, len(scores))
	for i, s := range scores {
		vals[i] = s.LEKScore
	}
	return poindexter.ComputeDistributionStats(vals)
}

// ComputeGrammarAxisStats returns per-axis distribution stats for grammar features.
func ComputeGrammarAxisStats(entries []ScoredEntry) []GrammarAxisStats {
	points := make([]poindexter.KDPoint[ScoredEntry], len(entries))
	for i, e := range entries {
		points[i] = poindexter.KDPoint[ScoredEntry]{
			ID:     e.ID,
			Coords: GrammarFeatures(e.Grammar),
			Value:  e,
		}
	}
	return poindexter.ComputeAxisDistributions(points, GrammarFeatureLabels())
}

// SummaryReport holds aggregate analytics for a scored population.
type SummaryReport struct {
	Total          int
	CompositeStats ScoreDistribution
	AxisStats      []GrammarAxisStats
}

// ScoreSummary computes a full analytics report from scored entries.
func ScoreSummary(entries []ScoredEntry) SummaryReport {
	scores := make([]GrammarScore, len(entries))
	for i, e := range entries {
		scores[i] = e.Grammar
	}
	return SummaryReport{
		Total:          len(entries),
		CompositeStats: ComputeScoreDistribution(scores),
		AxisStats:      ComputeGrammarAxisStats(entries),
	}
}