stat_glmm

A compact collection of Jupyter notebooks illustrating generalized linear mixed models (GLMMs) and related hierarchical modeling workflows using synthetic semiconductor manufacturing examples.[page:1]

Overview

This repository focuses on worked notebook examples rather than a packaged Python library. The notebooks cover binary outcomes, count outcomes, and continuous-response mixed models with random effects, using semiconductor yield, leakage, and defect-count scenarios as motivating examples.

Repository contents

Notebook	Summary
binomial_yield.ipynb	Builds a hierarchical binomial/logistic modeling workflow for semiconductor yield, starting from synthetic data, exploratory analysis, and pooled logistic regression, then moving to mixed-effects modeling for pass probability.
poisson_negbin_count.ipynb	Studies defect-count data using Poisson and negative-binomial models, including mixed-model structure and overdispersion handling for hierarchical count data.
random_intercept.ipynb	Introduces a random-intercept mixed model for leakage current as a function of humidity, with variance decomposition and comparison against a pooled linear baseline.
random_slope_inter.ipynb	Extends the mixed-model setup to include both random slopes and random intercepts, emphasizing humidity-specific variability and out-of-sample model comparison.

Themes covered

• Generalized linear mixed models (GLMMs) and hierarchical modeling.
• Synthetic semiconductor manufacturing datasets for yield, leakage, and defect analysis.
• Model comparison between pooled and mixed-effects approaches.
• Binomial, Poisson, and negative-binomial response modeling.
• Normal models with random intercept and random slope formulations.
• Practical notebook-based analysis with statsmodels, PyMC, numpy, pandas, matplotlib, and related scientific Python tools.

Intended use

This repo is best used as a small reference library of worked examples for learning or revisiting core GLMM patterns. It is especially useful for users who want notebook-first demonstrations of hierarchical modeling ideas in an applied setting rather than a production-ready software package.

Notes

The repository currently contains four notebooks and an Apache-2.0 license, but no README on the GitHub landing page.[page:1] Adding this summary would make the project easier to understand at a glance.[page:1]