does weka predictor work well?

hi all,
i’m wondering if there is’nt any troubles with weka predictor cause when i want to use this node , it always raise ERROR Weka Predictor Execute failed: (“ArrayIndexOutOfBoundsException”) exception.

Hi: I’m using weka predictor and works fine for me. Check the model and test data you’re feeding into it, cause if they don’t match you can get errors. They have to match in quantity and type of attributes, so look for that

However, this is very likely a bug, we need to fix. Do you have some more information on how to reproduce it. Thanks, Thomas


Not sure if this issue has been fixed, but i ran into it and figured out how to circumvent the problem, though it is a bug that probably should be fixed.

Typically a column filter must be provided before the weka learner to remove any column that should not be used. If one then tries to pass the original table (one that contains all of the column removed upstream of the weka learner), you will get this error. Now if the table to classify has the same columns as the one used for the learning step, no more issues.

This issue occurs in some cases with the Weka clustering nodes, but with a different error.

It would be good if any predictor/clusterAssigner nodes would not fail if more columns of data (or annotations) were provided, as compared to what is used by the learner/clustering nodes.