Pyglmnet: A Python package for elastic net regularized generalized linear model
summary
An open source Python package implements elastic net regularized generalized linear models as a practical alternative to R glmnet, covering Poisson regression and multinomial cases and producing sampled draws from fitted models. Optimization uses proximal updates with soft-thresholding, warm restarts over a regularization path, an active set and cyclic coordinate descent with optional Newton steps. Documentation, notebooks and GitHub code exist and benchmarks against scikit-learn are included, the package is usable now for GLMs with elastic net penalties.
AI-written summary · about
chapters (10)
- 0:00Pyglmnet and GLMs intro
- 3:39Why Pyglmnet was developed
- 7:58Linear regression to GLMs
- 14:40Optimizing GLM parameters
- 19:20Understanding regularization
- 26:44Elastic net: Ridge and Lasso
- 30:44Pyglmnet optimization tricks
- 37:30Symbolic and automatic differentiation
- 41:26Pyglmnet API usage
- 43:39Summary and next steps
related talks
32mpyGAM: balancing interpretability and predictive power using...
01h 30mA Primer on Gaussian Processes for Regression Analysis
02h 10mLasso and Ridge regression models in R: Introduction
39mBayesian optimisation with scikit-learn
31mLogistic Regression Behind the Scenes
42mBayesian Non-parametric Models for Data Science using PyMC3
23mparsnip A tidy model interface
01h 15mEducational framework for Black Box optimization methods design