R mlr Methods for defining parameters

Hi,

I have a question about R mlr Methods for defining parameters from a blog post, point “4. Random Forest”:

It is possible (let me know, if I’m wrong) to define parameters:

  1. rf <- makeLearner(…, par.vals = list(ntree = 200))
  2. rf$par.vals <- list(ntree = 200)
  3. rf_param <- makeParamSet( makeIntegerParam(“ntree”, lower = 50, upper = 500))

What are the differences between these methods? Does 1 and 2 are the same result, just different methods? Does 1 and 3 are the same, just different methods? And 2 and 3 ? Which of them is primary?

When I tune parameters, then I see, that the process go through from 50 to 500 (for example using Grid search). Then what is the impact of 1 and/or 2 ?

Please, help!

© Copyright 2013-2019 Analytics Vidhya