Filters out numeric columns, which have a low variance.
Applies a feature filter model built during feature selection
Splits the rows from the input table into rows that were successfully processed and rows that were not processed on the cluster.