FCBMA R package

Factor collapsing with Bayesian model averaging for regression models

Authors:

• Sen Hu (ethansen.hu@gmail.com)
• Adrian O’Hagan (adrian.ohagan@ucd.ie)
• Thomas Brendan Murphy (brendan.murphy@ucd.ie)

Description

The FCBMA package uses factor collapsing (FC) and Bayesian model averaging (BMA) to find the optimal manners of combinations of categorical levels within categorical variables (i.e. clulstering of categorical levels) in linear or generalized linear regression models, as introduced in Hu et al (2018)

Installation

You can install the latest development version of FCBMA from GitHub:

install.packages("devtools")
devtools::install_github("senhu/FCBMA")

Then the package can be loaded with:

library(FCBMA)

Example

This README file follows a package vignette format, and an example is briefly demonstrated using the Swedish third party motor insurance claims data in 1977 (available in this package), as illustrated in Hu et al (2018). The data can be loaded via

data("sweden")

Details about the data set can be found in the data set documentation within this package.

We start with the frequency aspect of claim modelling, by building a baseline model where categorical variables are unchanged:

freq <- glm(Claims ~ Zone + Bonus + Make + Kilometres, offset = log(Insured), data = sweden, family = "poisson")

Then the factors can be collpased individually, via

freq.kilo <- FCBMA(freq, varia.list = c("Kilometres"),
                   method = "complete", verbose = FALSE)
freq.make <- FCBMA(model = freq, varia.list = c("Make"),
                   method = "SA", verbose = FALSE)

or they can be collpased all together, via

freq.all <- FCBMA(freq, 
                  varia.list = c("Kilometres", "Zone", "Bonus", "Make"),
                  method = "GA", verbose = FALSE)

Depending on the size of the potential model space to search for the opitmal combinations of collapsing, either greedy/complete search or stochastic search with simulated annealing or genetic algorithm can be used, by setting method = "complete", "SA" or "GA".

Reference

Hu, S., O’Hagan, A., and Murphy, T. B. (2018) Motor insurance claim modelling with factor collapsing and Bayesian model averaging. Stat, 7: e180. doi: 10.1002/sta4.180.