As you can see, I’m facing a problem with H2O node when i try to include a column that has been log transformed which is: ERROR Table to H2O 0:333:78:331 Execute failed: Job crashed without obvious reason! Cause: DistributedException from localhost/127.0.0.1:55129: ‘null’, caused by java.lang.ArrayIndexOutOfBoundsException See log for details.
However, there’re other included columns with log transformation but they don’t cause the above error, it’s just that specific column which cause this error.
could you tell us more about the variable you are trying to use? Do they have an unusual format or NaN (not a number) included? The variable format is a double number (Salary)
How did you do the LOG transformation? What node was used? Column Expression Node.
What version of KNIME and the H2O nodes are you running? 4.3.2
How largest the dataset, can you try with a smaller one and check the memory settings of your KNIME installation. ~20K
If possible could you provide is with an example of the failing value? There’s no specific value specified in the error log.
@JayR you can try with the latest versions of KNIME and H2O but I doubt that that will make any difference.
Then you might want to narrow down the failing column (try to exclude some that might cause the problem and see how that goes) and again check for NaN and Memory like suggested.
Maybe provide us with a sample workflow and data that would fail.
