使用partykit创建随机森林时，如何获得变量的重要性？

Question

I'm using partykit to create a multi-output random forest. The library party provides a function (impVar) to obtain it directly, but I cannot find an equivalent function in partykit. Is there any way to obtain the variable importance to classify with partykit?

Answer 1

Not yet. While the tree fitting functions ctree() and mob() in partykit already supersede the corresponding functionality in the party package, the same is not yet true for cforest() . At the moment the latter is a basic re-implementation that has already some new features but also lacks various details, specifically methods for flexible prediction and variable importance. We plan to add these in the future, though.