Hello everyone,
My current task is to implement two separate workflows for a blind validation of several models. I would train several models (Gradient Boosted Trees, Random Forest, etc.) with Workflow 1 on a large dataset, and cache the models on the disk. My colleague would then read the saved models with Workflow 2, and validate them on a different dataset.
I saved the binary classification threshold optimised on the training set, as calculated by the BinaryClf inspector in Workflow 1. I would like to overwrite the thresholds in Workflow 2 with these, but I’m not sure how.
There seems to be a flow variable option (inViewThresholdValues) in the BinaryClf inspector dialogue box where I can define different thresholds for different models:
And I imagine this must be the expected format for the said flow variables:
However, defining this flow variable doesn’t seem to have any effect. It seems I can only define one uniform custom threshold for all models in the “Initial threshold methods” box shown here: (This is clearly suboptimal for me)
