Cross-validation and feature selection

[Sliktrav]

Hey, So I’m trying to develop a workflow that allows me to do 5-fold cross-validation but with each iteration do a feature selection and at the end, compile all of the best features from all of the 5-fold cross-validation to be used to build a model

[armingrudd]

Hi @Sliktrav and welcome to the KNIME Community Forum,

For cross validation you can use these nodes:

X-Partitioner

X-Aggregator

And for feature selection:

Feature Selection Loop Start (1:1)

Feature Selection Loop End

In case you want to do parameter optimization as well:

Parameter Optimization Loop Start

Parameter Optimization Loop End

You can find workflow examples for each node in their correspoinding KNIME Hub webpage as well. In case you have further questions, don’t hesitate to ask.