Github user MrBago commented on the issue: https://github.com/apache/spark/pull/20566 I believe this will break persistence for LogisticRegression. I believe the issue is that the `threshold` param on LogisticRegressionModel doesn't get a default directly, but only gets it during the call to `fit` on LogisticRegression. This is currently fine because the Model can only be created by fitting or by being read from disk and in both case some value gets set for threshold. With this change that's no longer the case. Here's a test to confirm, https://github.com/apache/spark/commit/5db2108224accdf848b41ef0d8d1c312b49f49c6. I believe LinearRegression may have a similar issue. Our current tests don't seem to cover this kind of thing so I think we should improve test coverage if we want to make this kind of change.
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org