trappmartin / bayesiansumproductnetworks Goto Github PK

From the installation instructions, I assume that these used to exist? I've looked through the module files and there clearly are dependencies :-) While I can easily populate the dependencies for a pull request, I'm less sure about filling in the other fields.

I was digging around in the preprocessing code and found, that there are two heuristics for identifying discrete variables. One checking for if all nonmissing values are integers L199 and the other one counting unique values L200. Which one did you use in the experiment and also maybe why do you prefer one over the other?

One other thing I noticed is, that casting of binary variables into 0,1 values is applied to the whole dataset, not just the particular column L212 , fortunately the heterogeneous data from ABDA paper should have the binaries removed. However this is not the case with Anneal-U dataset, which preprocesses the 9th feature as categorical instead of binary, as I believe that the authors probably used the value range metadata. As a result there is some discrepancy caused by the different "detection" of discrete variables, but I cannot judge if it has some negative effects on the results in that particular dataset.

@trappmartin I generated a small continuous synthetic dataset for familiarizing myself with your package and the following issue occurred: All my continuous values are replaced with integers in run_abda_rg.jl. I believe the culprit is

 dids = filter(d -> length(filter(x -> !isnan(x), unique(x_train[:,d]))) < args["maxcategories"], 1:D)

which really should read

 dids = filter(d -> length(filter(x -> !isnan(x), unique(x_train[:,d]))) < args["maxcategories"], dids)