casact / rp-bnn-claims

Individual Claims Forecasting with Bayesian Mixture Density Networks

bnn-claims

Repository for the paper "Individual Claims Forecasting with Bayesian Mixture Density Networks." This work is supported by the Casualty Actuarial Society.

Data

The data files used in the code are part of the release. They can be downloaded after cloning the repo by running

# remotes::install_github("ropensci/piggyback")
piggyback::pb_download()

The raw data file Simulated.Cashflow.txt was created using the simulation machine available at https://people.math.ethz.ch/~wueth/simulation.html.

Dependencies

(GPU only) To install the necessary dependencies, install the latest dev version of renv using

remotes::install_github("rstudio/renv")

then run renv::init() followed by renv::restore().