A one-class classification decision tree based on kernel density estimation

Sarah Itani, Fabian Lecron, Philippe Fortemps

Abstract: Highlights•One-Class Classification (OCC) addresses the challenging issue of class unbalance.•OCC models are trained on the instances of a class and some few potential outliers.•A new One-Class Tree (OC-Tree) is proposed for explainable and accurate decisions.•The tree induction is driven by density estimation to isolate target groupings.•The model proved efficient to diagnose ADHD and is promising for clinical practice.