fyp-statistical-validator

Confidence interval calculation automates computation of 95% CIs using multiple methods including normal approximation, bootstrap, and multi-seed approaches. This Skill provides ready-to-use Python APIs that compute intervals across model predictions and experimental runs, returning thesis-ready statistics without manual computation.

Hypothesis testing suite performs McNemar's test, paired t-test, and Bonferroni correction for multiple comparisons. The Skill automates test execution across model pairs and seeds, generating statistical significance results formatted for thesis chapters.

APA table generation converts raw experiment statistics into publication-ready tables following APA style conventions. The Skill formats confidence intervals, test results, and sample sizes automatically, eliminating manual table construction for thesis submission.

Reproducibility documentation generates standardized statements capturing random seeds, model versions, and data splits used in experiments. The Skill creates thesis-ready reproducibility sections that satisfy TAR UMT standards across multiple experimental runs.

Yes. The Skill integrates with numpy, scipy, scikit-learn, statsmodels, and pandas to extract predictions and compute statistics. It accepts outputs from these libraries directly, automating validation workflows without data reformatting.

Statistical validation covers 95% confidence intervals, hypothesis tests, Bonferroni correction, and APA table formatting per TAR UMT thesis standards. The Skill validates ML experiment results against these academic requirements end-to-end.