Abstract

In Random Forests, proximity distances are a metric representation of datainto decision space. By observing how changes in input map to the movement ofinstances in this space we are able to determine the independent contributionof each feature to the decision-making process. For binary feature vectors,this process is fully specified. As these changes in input move particularinstances nearer to the in-group or out-group, the independent contribution ofeach feature can be uncovered. Using this technique, we are able to calculatethe contribution of each feature in determining how black-box decisions weremade. This allows explication of the decision-making process, audit of theclassifier, and post-hoc analysis of errors in classification.

 

Quick Read (beta)