jreps

The current lasso logistic regression we use has group lasso but it is still being tested. Hopefully it will be added this year.

Hi, I tested integrating Spark a couple of years ago and found that the overhead to set up the parallelization wasn't worth it with the data size I was using....

I think, currently the validation loads every model and then saves the results to the Validation/data/developmentModelAnalysisName folder. The assumption was external validation uses the same settings as the development data,...

I like it! Much easier and fixes the issue. Do you want to make the edits and make a pull request or shall I make the edits?

I recently added an option to ignore the covariateSummary to speed things up if not needed. I tried to add indexes but I'm not sure I optimized them - I'm...

Martijn - do you think it is possible to use the Andromeda parallel code by covariateId and then convert to a data.table? Not sure that is possible for joins though...

I actually made edits to speed this up a lot in the development branch. Not sure I updated in the master though. I've now merged development into master so that...

Are you still getting speed issues @tomseinen ? I updated the code and added indexes, so hopefully it is faster?

I found the restriction to the population was taking a long time and the inner join did not scale (even with indexes) - I've now modified that. The summary still...

Hello. This error is expected, but we should have an input check to warn people and clearer documentation. The majority of the PLP classifiers only support 2d (patient x covariate)...