This time we want to explore how the uncertainty estimation of random forests behaves under different settings.
Create the learner:
In our toy example we cut some data to see how the interpolation and the uncertainty behaves in that gaps.
With a little help of the ParamHelpers Package we can easily generate an exhaustive set of different settings:
Now we can iterate over all different parameter settings and see how it turns out.
That all doesn’t look so reliable to me.