Instance-dependent cost-sensitive learning for detecting transfer fraud

Card transaction fraud is a growing problem affecting card holders worldwide. Financial institutions increasingly rely upon data-driven methods for developing detection systems, which are able to automatically detect and block fraudulent transactions. From machine learning perspective, the task of detecting transactions binary classification problem. Classification models commonly trained evaluated in terms statistical performance measures, such as likelihood AUC, respectively. These however, do not take into account actual business objective, minimize financial losses due fraud. Fraud be acknowledged an instance-dependent cost-sensitive problem, where costs misclassification vary between instances, requiring adapted approaches model. In this article, threshold derived, based on cost matrix transfer detection, that allows making optimal cost-based decision each transaction. Two novel classifiers presented, lasso-regularized logistic regression gradient tree boosting, directly proposed measure when The implemented R packages cslogit csboost, compared against state-of-the-art publicly available data set from competition website Kaggle proprietary set. results experiments highlight potential reducing by adopting methods.