Statisticsmath
The mathematics of collecting, summarizing, and inferring from data — distributions, means, variance, hypothesis testing.
Evaluating compiler output (coverage, confidence distributions, misconception frequency) is statistical; understanding it prevents over-trusting single runs.
Where it appears
- SKCE build report (coverage
- model vs deterministic split)
Descend
Historical evolution
From Fisher and Pearson to modern ML evaluation metrics.
Implementation details
SKCE reports coverage as a ratio and model-vs-deterministic node/edge counts in the manifest.
Sources
- [compiler/artifact_store.py] compiler/artifact_store.py
- [compiler/artifact_store.py] compiler/artifact_store.py