Bank Account Fraud (BAF)
收藏资源简介:
Bank Account Fraud (BAF) 数据集是由 Feedzai 和 Porto 大学科学学院合作创建的,旨在为机器学习研究提供一个大规模、隐私保护的现实表格数据集。该数据集通过应用最新的表格数据生成技术,基于一个匿名的真实世界银行账户开户欺诈检测数据集生成。数据集包含六种不同的变体,每种变体都包含了特定类型的数据偏差,以允许实践者测试机器学习方法的性能和公平性。这些数据集的应用领域主要集中在金融服务的欺诈检测,旨在解决在动态环境中评估新旧机器学习方法的性能和公平性的问题。
Bank Account Fraud (BAF) dataset was co-developed by Feedzai and the Faculty of Science of the University of Porto, aiming to provide a large-scale, privacy-preserving real-world tabular dataset for machine learning research. This dataset is generated based on an anonymized real-world bank account opening fraud detection dataset using state-of-the-art tabular data generation techniques. It includes six distinct variants, each containing specific types of data bias, to enable practitioners to test the performance and fairness of machine learning methods. These datasets are mainly applied in the field of financial service fraud detection, targeting the problem of evaluating the performance and fairness of both new and existing machine learning methods in dynamic environments.




