Implementing BTYD Models with the Log Sum Exp Patch

Functions for data preparation, parameter estimation, scoring, and plotting for the BG/BB (Fader, Hardie, and Shang 2010 ), BG/NBD (Fader, Hardie, and Lee 2005 ) and Pareto/NBD and Gamma/Gamma (Fader, Hardie, and Lee 2005 ) models.

BTYD 2.4.2

A patch for the pnbd.LL() function in the original BTYD package, first proposed by Theo Strinopoulos. Now on CRAN, so you can get it the usual way, with install.packages().

Justification

In its original version the Pareto/NBD (pnbd) part of the BTYD package failed for me, as it did for Theo, for reasons he explained here. So, I implemented his fix, rebuilt from source, and then it worked. This is the short version. The long version is that in the process of implementing the fix I made some changes to the choice of optimization routine (now using optimx as opposed to optim), Gaussian hypergeometric (now you have the option of using the hypergeo package) and I found some opportunities for refactoring functions defined in both the pnbd and the bgnbd (Beta-Geometric Negative Binomial) groups. I explained these changes are explained in separate documentation (fix_pnbd.html). Finally, I rebuilt BTYD and checked that it would return the same numbers as BTYD 2.4 did when running the examples in the original BTYD vignette (see threeway_walkthrough.R).

How to install from here

First git clone.

If you use devtools, then at the R console just call document(); build(); install(build_vignettes = TRUE). Done.

Otherwise, build the standard way in two steps:

1. At the command line, build and check the source tarball:

R CMD build BTYD
R CMD check BTYD_2.4.2.tar.gz

1. Then, at the R console, install it:

install.packages("BTYD2_2.4.2.tar.gz", repos = NULL, type = "source")
# works only in RStudio:
.rs.restartR()