Use & relation of weightCol in Classifiers and MulticlassClassificationEvaluator

110 Views Asked by At

I'm learning pyspark and am currently working on an imbalanced dataset which I want to use in a classifier. Most ml classifiers support a weightCol parameter in which I can set weights for my observations. I also noticed that the MulticlassClassificationEvaluator also has a weightCol parameter. I intend to use the MulticlassClassificationEvaluator to evaluate my model.

How does weightCol in a classifier relate to the weightCol in MulticlassClassificationEvaluator? If I use weights in my classifier, am I supposed to pass the same column in MulticlassClassificationEvaluator when I want to obtain accuracy or other metrics?

0

There are 0 best solutions below