The Statistics Notebook

Jenny Lee¶

Study notes on statistical learning and experimental design, developed from lectures at the University of Waterloo.

Evaluation, preprocessing, and the concepts used throughout the book.

Logistic regression, nearest neighbors, Naive Bayes, trees, ensembles, and support vector machines.

Prediction error, model assessment, regularization, splines, LASSO, and regression trees.

Treatment comparisons, blocking, factorial designs, response surfaces, and hierarchical models.

The lectures form the basis of these notes; the organization and supplementary explanations are my own.

A Statistical Framework for LLM-as-a-Judge Metrics

Reliability, self-preference and metric redundancy.

Numbered records of local model experiments and their results.

Dataset guides and viewers for research, including MTRAG.