Forest Tool Output - Need Feature Importances

[jnell2]

I built a random forest using the Forest Model tool. The output gives me a variable importance plot according to Mean Decrease Gini. This is helpful, but I need the actual Feature Importance Percentage. Something like the following, where the percentages all add up to 100%: 

Variable 1 = 40% 
Variable 2 = 20%

Variable 3 = 15%

Variable 4 = 10%

Variable 5 = 10%

Variable 6 = 5%

I suspect I'll have to change the Forest Model tool, but I'm not familiar with R. If anyone could help me, it would be very much appreciated. 

[SophiaF]

Giving this a bump in the hopes that some of our R experts can give some advice - also, I would suggest posting this to Product Ideas if you think others would find it useful!

[TimothyL]

Hi @jnell2 , 

We recently published a new set of model explanation macros: LIME & SHAP in this post.

It supports Forest Model and other predictive tools including Gradient Boost, SVM, and more!. Give a try and see if it could help you!