Help files/examples

[hjuerges]

Thanks for fixing the bug I reported earlier. This really helps using ddml in my "causal inference" course.

However, there is another (new?) issue:

When running the example from the help file:

# Estimate the partially linear model using short-stacking with base learners
#     ols, lasso, and ridge. We can also use custom_ensemble_weights
#     to estimate the ATE using every individual base learner.
weights_everylearner <- diag(1, 3)
colnames(weights_everylearner) <- c("mdl:ols", "mdl:lasso", "mdl:ridge")
plm_fit <- ddml_plm(y, D, X,
                    learners = list(list(fun = ols),
                                    list(fun = mdl_glmnet),
                                    list(fun = mdl_glmnet,
                                         args = list(alpha = 0))),
                    ensemble_type = 'nnls',
                    custom_ensemble_weights = weights_everylearner,
                    shortstack = TRUE,
                    sample_folds = 2,
                    silent = TRUE)

I get the following error message:

Error in (function (what, args, quote = FALSE, envir = parent.frame()) : 'what' must be a function or character string
This happens everytime I use more than one learner.

[thomaswiemann]

@hjuerges the code in the examples works on my end---can you check whether running the full example, e.g. for ddml_plm, works?

I suspect that your issues stem from a quirk in the current ensemble implementation: When specifying a single learner, the argument uses what to pass the learner:

learners = list(what = mdl_glmnet,
                args = list(alpha = 0))

but when specifying several learners it instead uses fun:

learners = list(list(fun = ols),
                list(fun = mdl_glmnet),
                list(fun = mdl_glmnet,
                     args = list(alpha = 0))),

There's no good reason not to use either what or fun throughout. Will fix this in a future release.

[hjuerges]

Thanks for looking into this.

Yes, the fun versus what is a bit inconvenient, but that was not my problem. I know how that sounds, but I ran exactly your code example from within the help window in RStudio when I got that message.

Now here comes the important part: I don't understand why, but works now.