Welcome to glum’s documentation!

glum is a fast, modern, Python-first GLM estimation library. Generalized linear modeling (GLM) is a core statistical tool that includes many common methods like least-squares regression, Poisson regression and logistic regression as special cases. In addition to fitting basic GLMs, glum supports a wide range of features. These include:

  • Built-in cross validation for optimal regularization, efficiently exploiting a “regularization path”

  • L1 and elastic net regularization, which produce sparse and easily interpretable solutions

  • L2 regularization, including variable matrix-valued (Tikhonov) penalties, which are useful in modeling correlated effects

  • Normal, Poisson, logistic, gamma, and Tweedie distributions, plus varied and customizable link functions

  • Built-in formula-based model specification using formulaic

  • Classical statistical inference for unregularized models using dispersion and standard errors

  • Box and linear inequality constraints, sample weights, offsets

  • A scikit-learn-like API to fit smoothly into existing workflows

glum was also built with performance in mind. The following figure shows the runtime of a realistic example using an insurance dataset. For more details and other benchmarks, see the Benchmarks section.

_images/wide-insurance-gamma-normalized.png

We suggest visiting the Installation and Getting Started sections first.

Index