Hi, I set up a problem using the workflow suggested at Random Forest, Gradient Boosted Trees, and TreeEnsemble – KNIME Community Hub . I haven’t been able to assign weights to the classes that represent adequately the costs of false positives, though. For example, let’s say that the cost of a false positive is 4x the value of a true positive. How can I set up the workflow to take that into account? What I have been doing so far is a lot of trial-and-error in Excel to approximate the best cutoff value (e.g., accepting only the cases in which the class attributed to the item has a prediction value >.8), but I would like to do it in a more rigorous way. Thanks!

Applying weights to Random Forest Prediction

Mpattadkal November 7, 2023, 3:17pm 2

Please have a look at this threads

Let me know if the discussion here solves your query.

Thank you!

1 Like