[pyspark] eval_result is not working

[wbo4958]

    @wbo4958 @WeichenXu123 

Hi guys, sorry for my late response to this thread. I finally had the chance to try the new code and it works well!!

I just had one problem when I am trying to get the training loss by using this:

results = {}
xgb_classifier = SparkXGBClassifier(**xgb_parms, validation_indicator_col='isVal', evals_result=results, verbose_eval=25, num_workers=16, n_estimators=250, early_stopping_rounds=25, label_col=name)

Then I got this error:

TypeError: train() got multiple values for keyword argument 'evals_result'

It seems that the evals_result is already set within SparkXGBClassifier, but then how can I retrieve the results? I couldn't find anything in the documentation, then I tried some options myself, but none of them works. Thanks!

Originally posted by @faaany in #8221 (comment)

[wbo4958]

I will fix it.