This post presents a function mlReserve() for machine-learning loss reserving on ‘real-world’ triangles, based on a fork of the ‘ChainLadder’ package. It includes two experiments, each scored against a known true reserve: (A) Two triangles from an individual-claims simulator (Wang & Wüthrich), embedded below: chain ladder vs mlReserve, in detail; (…

Pipeline setup
{pipeflow} builds a pipeline by adding one R function per step with
pip_add (see also the Get
started article). In this post, we’ll work with the following toy
pipeline:

step load produces a sequence
step clean doubles it
step fit sums it, and
step report formats …