I’m a University Student and for my subject Data Science I have to use knime to solve a kaggle problem. Therefore I’m very inexperienced using knime.
It’s a group project and my part is using a neural network for solving the problem.
My problem is that in the end the Scorer node will stop at 50% with the Error “Execute failed: Java heap space”.
After consulting google the only solution I found was increase the memory space.
But after increasing it from 2GB to 10 GB I still get the same problem.
It seems strange that the Scorer mode would exhibit this kind of failure. My first thought is that maybe you’re not using the appropriate columns for scoring? Generally you’re going to use (for example) Target in your first column and Prediction (Target) in your second column. If you’re instead including confidence or probabilities you could end up with a large matrix which might cause memory problems.
Can you upload the current version of your workflow so someone can take a look?
thanks for your fast reply.
I exported the workflow and I hope I got the Upload right.
I’m not sure if it works right with just the workflow because we use a dictionary with the String Replacer and I’m not sure if this is included.
I could upload all data but since were all inexperienced in KNIME it’s not a really nice solution so there are 43 String Replacer in a row, all of which are using a dictionary
But here are the Screenshots of the Scorer.
I hope that could help
