On the use of case estimate and transactional payment data in neural networks for individual loss reserving

This repository contains the code and datasets used to produce the results in the paper "On the use of case estimate and transactional payment data in neural networks for individual loss reserving". This repository can also be found on GitHub.

All 5 zip files should be downloaded, extracted and combined into a single folder.

The files should be examined and run in the following order:

1. Generate Dataset.R

As the name suggests, this file is responsible for simulating the datasets from SPLICE (Avanzi, Taylor & Wang, 2023) and SynthETIC (Avanzi, Taylor, Wang & Wong, 2020).

2. Data Manipulation.R

Contains the main data manipulation, as well as train-test splitting. Prepares the raw data for input into the RNN(+) and FNN(+) models.

3. Model Training.ipynb files

These jupyter notebooks rely on 'Functions.py'. This script contains all the functions and classes to be called from each of the model training notebooks.