It would make sense to compare our results for the semiparametric estimation approach (particularly the mte results) with the ones obtained from competing R packages. For this purpose I would suggest the following procedure:
Simulate (lets say 100) data sets with our simulation routine (and save the associated init files in a json file)
Write an R script that computes the MTE for each of this simulated samples and save the results in a json file
Include a regression test in our current test frame work in which we estimate the mte with grmpy and compare the resulting mte with the one obtained from our R script
It would make sense to compare our results for the semiparametric estimation approach (particularly the mte results) with the ones obtained from competing R packages. For this purpose I would suggest the following procedure:
Simulate (lets say 100) data sets with our simulation routine (and save the associated init files in a json file)
Write an R script that computes the MTE for each of this simulated samples and save the results in a json file
Include a regression test in our current test frame work in which we estimate the mte with grmpy and compare the resulting mte with the one obtained from our R script