Three questions...

[richha...@gmail.com]

I did an initial set of runs using your package.  [I fixed impute.funct ] to do that.

In the small chance the authors are still doing updates now and then, may I ask three questions:

1.  I seem to be getting Excel tables and other results with variables names v1, v2, etc.  Is there an error I am making that results in this.  Alternatively, is there a quick change in the code I can make so that this does not occur?

2.  Are there any more recent updates that make use of additional regression functions in R?

3.  Has anyone done a version that allows for different runs to be made on different CPUS, where results are combined?  As I see it, this would involve ( in Step #9 )  having the data frames dfMod and dfRes created from output files in a function that could be run separately.

Thank you very much

Rich Haney

[muntisa]

Hello,

1 - The change of the feature names is made in the first steps of RRegrs. There is no error and it is due to the caret processing of the dataset. Of course it could be changed in the future.

2 - Nope, we didnt included more regression methods in RRegrs.

3 - Yes, we created batchRRegrs (https://github.com/cafernandezlo/batchRRegrs) in order to use command lines for our HPC cluster. Thus, you can run ONE method only, you need to make the splits, you need to compare the results between the methods, you dont have the PDF outputs, etc. The advantages: the model is saved as R object, so you can load it and create any plots or statistics you consider. In addition, you can easly modify any parameter of the methods such as the number of neurons or the number of the features to feed each tree for RF method, etc.

For the moment, I think RRegrs is not evolving, so I an not sure when we will add any new functionality.

Thanks for using RRegrs!

Cristian R Munteanu

UDC, Spain