Technische Universität Berlin, Germany
Despite the large success of Random Forests, there barely exist suitable visualizations, which allow a fast and accurate understanding of how well they perform a certain task and what leads to this performance. This work proposes an exemplar-driven visualization illustrating the most important key concepts of a Random Forest classifier, namely strength and correlation of the individual trees as well as strength of the whole forest. A visual inspection of the results enables not only an easy performance evaluation but also provides further insights why this performance was achieved and how parameters of the underlying Random Forest should be changed in order to further improve the performance.
VRF - Visual Random Forests
===
VRF provides a Matlab script, which visualizes the data of a given RF-classifier by three-dimensional plots. All important properties of an RF are easily comprehensible from the rendered 3D space (ie. strength of the whole forest as well as of individual trees, correlation of the trees, leaf impurity, and split dimensions).
Besides Matlab, no other external dependencies are required.
R. Hänsch, O. Hellwich, Performance Assessment and Interpretation of Random Forests by Three-Dimensional Visualizations, IVAPP 2015 - Proceedings of the International Conference on Information Visualization Theory and Application, March 2015
@INPROCEEDING{ Haensch:2015:IVAPP15, author = {Hänsch, Ronny and Hellwich, Olaf}, title = {Performance Assessment and Interpretation of Random Forests by Three-Dimensional Visualizations}, booktitle = {Proceedings of the International Conference on Information Visualization Theory and Application, IVAPP 2015}, volume = {tba}, pages = {tba}, year = {2015} }