I want to compare different values for k in a nearest neighbor problem, using the parameter optimization loop nodes.
When I use the “regular” Scorer Node (two out ports) I can pick accuracy or Cohen’s kappa as the value to optimize in the loop end note (connecting the scorer and the end node by the flow variable ports).
However, when I replace the Scorer Node with the Scorer (Java Script) Node, it seems not to provide the needed flow variables for the loop end node any longer.
What am I missing? Is there a way to generate the flow variables in the Scorer Java Script node?
That information isn’t publicly available - we track it on our internal Jira system. Feel free to ask here on the forum if you ever run into something that is confusing or seems odd