tess-st/shinyBMR: A graphical user interface to explore benchmark results and make blackbox methods interpretable.

shinyBMR provides an user-interface for exploring the huge output from a benchmark study based on the benchmark() function out of the mlr-package. Due to practical experience these analyses mainly focus on classification tasks, where non-smoting and smoting (for imbalance correction) techniques in combination with non-tuning and tuning is used for the learners competing in the benchmark study. Of course one can also use this interface for analyzing regression tasks. But it has to be mentioned, that functionalities not go much further As complexe machine learning methods - refered to as blackbox methods - tend to win these kind of competitions the app also provides a framework for accecssing interpretability of these methods. This implementation is based on the additional mlr-package called 'iml'. Benchmark analyses and the use of tools for interpretation of blackbox methods are not therefore could be used independently of each other.

install.packages("remotes")
remotes::install_github("tess-st/shinyBMR")