Does anyone have any example resources for how to write good unit tests that specifically test the ML pipeline and algorithm training / testing?
There are lots of resources for testing e.g. testthat
, assertr
and even for mocking - including mockery
and many of these have great examples that make sense in isolation. In reality however, more complete coverage for testing an ML pipeline might involve combinations of the previously mentioned packages.
Any insights on using test doubles in R and some examples of how someone has used these approaches to unit test model pipelines would be fantastic.
Thanks!