I get this error message when trying to execute my workflow: “Execute failed: Input table’s structure differs from reference (first iteration) table: different column counts”
This only occurs when I tick “Append columns with normalized class distribution” in the Decision Tree Predictor. If I don’t click this option it works perfectly fine. I leave all the other settings unchanged.
How can I solve this and what is the reason behind this error?
I am really looking forward to your help!
The error message tells you what’s wrong. Your input table contains a different number of columns compared to the result of the first iteration.
As you say in your post, when you tell the decision tree predictor node to add a column, then error appears. When you disable this option, the error goes away.
It’s all consistent.
If you’re doing cross-partitioning, why do you need to add the class distributions?
I have the same problem. I need to create the distribution for the ranking of the predictions. Is there any way to work around the message?
Maybe you could write them to a separate file and then remove them before using the closing node for the Crossvalidation
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.