corepointer commented on pull request #1183: URL: https://github.com/apache/systemds/pull/1183#issuecomment-789310393
Testing training and validation would be out of scope of a unit test imho. And you are right, we don't want a lot of extra data files in the repository for testing. I'd merge the PR to staging with minor formatting changes. A few questions came to mind while looking over it: * What is sherlockPreprocessing.dml for and where is it used? That one also contains a path reference to a non existent directory. * How did you test the implementation? And where did you get the test data from? What about a simple shell script that downloads the needed data (if that exists somewhere online) and fires off the algorithm. Regards, Mark ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
