作者: Thomas W. Yee
关键词: Dimensionality reduction 、 Statistics 、 Covariate 、 Generalized linear model 、 Scoring algorithm 、 Multinomial logistic regression 、 Multinomial distribution 、 Additive model 、 Iteratively reweighted least squares 、 Econometrics 、 Mathematics
摘要: Classical categorical regression models such as the multinomial logit and proportional odds are shown to be readily handled by vector generalized linear additive model (VGLM/VGAM) framework. Additionally, there natural extensions, reduced-rank VGLMs for dimension reduction, allowing covariates that have values specific each linear/additive predictor, e.g., consumer choice modeling. This article describes some of framework behind VGAM R package, its usage implementation details.