OHDSI / DeepPatientLevelPrediction

An R package for performing patient level prediction using deep learning in an observational database in the OMOP Common Data Model.
https://ohdsi.github.io/DeepPatientLevelPrediction
10 stars 4 forks source link

Empty test for training-cache #79

Closed egillax closed 1 year ago

egillax commented 1 year ago

Getting the following when running the tests:

  ══ Skipped tests (1) ═══════════════════════════════════════════════════════════
  • empty test (1): 'test-TrainingCache.R:49:1'

Its this test

The test is doing a runPlp where it should resume (or skip?) running hyperparameter tuning because of a previous run in same folder. But it's missing some kind of test of the output using some expect_* function from testthat.

Not sure if it can be detected from the output that it resumed correctly?

Another way to test would be to manually remove the last hyperparameter combination result from the ParameterPersistance.rds file and then run and verify it's updated.

@lhjohn This is coming from your test, but not high priority to fix

lhjohn commented 1 year ago

Updated the test to do as follows:

  1. Create empty cache on disk. It should return 1 as next param search index to assess.
  2. Run model on this existing cache.
  3. Ensure that cache now returns NA as next index to assess, which indicates that param search is complete.