01.03.2021 Add method summary that returns the number and proportion of appearance of each variables on a subset of selected model. The selection procedure proceed in two steps, first we select a dimension in which the mean, min or median is the lowest. Then we compute the selected percentile of the CV error on this dimension. We then select all models in all explored dimensions that have a lower CV error than the CV value set.
29.02.2021 Added support for regression problem. Changed the function swag to accept "Accuracy" as training loss function and specify the workflow to latter accept other loss function by specifying a procedure based on this loss function (see code of swag). Changed the function auto_swagControl to have different behavior depending on the type of procedure.
First version for submission to CRAN. The package is made available on GitHub (github.com/SMAC-Group/SWAG-R-Package/)