Section: New Results

Tree-based Cost-Sensitive Methods for Fraud Detection in Imbalanced Data

Participants: G. Metzler, X. Badiche, B. Belkasmi, E. Fromont, A. Habrard, M. Sebban

Bank fraud detection is a difficult classification problem where the number of frauds is much smaller than the number of genuine transactions. The authors of [20] present cost sensitive tree-based learning strategies applied in this context of highly imbalanced data. The paper first proposes a cost sensitive splitting criterion for decision trees that takes into account the cost of each transaction. Then the criterion is extended with a decision rule for classification with tree ensembles. The authors then propose a new cost-sensitive loss for gradient boosting. Both methods have been shown to be particularly relevant in the context of imbalanced data. Experiments on a proprietary dataset of bank fraud detection in retail transactions show that the presented cost sensitive algorithms increase the retailer's benefits by 1,43% compared to non cost-sensitive ones and that the gradient boosting approach outperforms all its competitors.