statistical-analysis

The skill determines appropriate hypothesis testing by applying normality, variance, and outlier diagnostics to your data distribution. It then recommends parametric or nonparametric tests based on the assumption check results.

The best way to report uncertainty is by computing effect sizes and confidence intervals across seeds. The skill provides structured reporting utilities to transparently document variance and statistical significance in CS experiments.

Use the Python utilities in scripts/ to compute power analyses for offline evaluations. The skill calculates required sample sizes and effect sizes to ensure your benchmarks and user studies have sufficient statistical power.

Yes, the skill integrates pandas and SciPy to perform normality, variance, and outlier diagnostics. You can run comprehensive_assumption_check directly on your dataset to obtain a full diagnostic summary.

Avoid parametric tests when the skill's assumption checks indicate non-normal distributions or unequal variances. In these constraint cases, the skill applies nonparametric alternatives for accurate ablation analysis.