Creating random forest after selecting feature of variable importance in MATLAB

Question

Is it possible to make feature selection of variable importance and then create a random forest in MATLAB?

I am using TreeBagger() with OOBPermutedVarDeltaError() to get the result of important features. But now I want to use these important features to recreate a random forest. Is it possible?

Answer 1

Use 50% of the data for parameter and feature selection. After this, then do 10-fold cross validation on the full data and check the performance. Then create a random forest model using full data for only training.