Predictor importance using Gradient Boosted algorithm

Hi ,
what are the options to Identify the predictor importance using Gradient Boosted algorithm?
I cant find node for that and in Python it’s quite common function.

Thanks a lot

Hi @Einavtam,

you can use the H2O Gradient Boosting Machine Leaner node for that. It outputs, besides the learned model, also a table showing the variable importance measures. To use this node you need to install the H2O Machine Learning Integration and convert your table to an H2O frame using the H2O Local Context and Table to H2O nodes. If you want to use the learned model to a prediction, the corresponding H2O Predictor node needs to be used.

I hope this helps you.

Cheers,

Simon

3 Likes

Thanks for your helpful answer .

It is possible to add extensions offline?

Yes, you could download the update site containing the extension as a .zip file and set is as an Update Site in the KNIME preferences. A more detailed explanation can be found here: https://www.knime.com/downloads/update

Cheers,

Simon