Feature selection | sequential search heuristic measure


I have a dataset with 33 features (F1-F33) and a class label. I then calculate the feature-class correlations and the feature-feature correlations using the rank correlation node. Then I calculate a metric using math formula and group by. This is calculated on the entire dataset though. I want to loop through and find the best combination of features that yield the maximum metric. I used feature selection loop as a start, but I cannot use a math formula as the inport to the feature selection loop end. Some sort of a sequential search would be best. Any tips would be greatly appreciated!

Hello @LizP ,

Welcome to the forum!

From the screenshot you provided, I can see the variable output of the ‘Table Row to Variable’ node is connected to the variable input of the Feature Selection Loop End node. Can you try connecting the input and output ports instead of the variable ports as shown below and try to execute the workflow?


If the issue persists, can you please share a KNIME workflow here with some dummy data, it would be helpful to understand the issue in detail.


1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.