README

With this git repo the simulations and figures from Kronziel et al. "Improving Random Forest Interpretability with Clustering-Based Ensembles of Artificial Representative Trees" can be reproduced. Please note, that the simulations in the paper were performed using a high throughout batch system. This repo will implement the same calculations on your local system, which may lead to a high computation time. 

To start, please first open simulations.R in RStudio. By running this script you should be able to generate a data set. The preselected settings will need around 10 minutes to calculate. Please note, that the calculations as in the paper will need more than 100 days on a local system. 

The scripts figure1.R, figure2.R and figure3.R can be used to reproduce the figures 1, 2 and 3 from the paper. You can choose if you want to build the figures from you self calculated data or to import the precalculated data from the paper, which you can find in the proc folder. 

For any issues or questions please contact me by l.kronziel[at]uni-luebeck.de.