Customer-level Fraud Detection Benchmark (CFDB)

客户级欺诈检测基准（CFDB）是一个专为提升机器学习模型研究和评估而设计的结构化数据集。该数据集由IBM开发，包含来自SAML-D和AML-World的详细客户行为和交易历史，总计约2076455条记录。CFDB通过将交易级数据转换为客户中心框架，强调客户行为模式，以更准确地模拟潜在欺诈。此数据集不仅遵守严格的隐私指南，确保用户保密，还通过提供丰富的客户中心特征信息，支持深度学习和异常检测算法等高级分析技术的发展。CFDB的应用领域主要集中在提升欺诈检测系统的预测准确性，促进金融安全领域的创新和改进。

The Customer-Level Fraud Detection Benchmark (CFDB) is a structured dataset specifically designed for the research and evaluation of machine learning models. Developed by IBM, this dataset contains detailed customer behavior and transaction history from SAML-D and AML-World, with a total of approximately 2,076,455 records. CFDB converts transaction-level data into a customer-centric framework, emphasizing customer behavior patterns to more accurately simulate potential fraud. This dataset not only complies with strict privacy guidelines to ensure user confidentiality, but also supports the development of advanced analytical technologies such as deep learning and anomaly detection algorithms by providing rich customer-centric feature information. The application scenarios of CFDB mainly focus on improving the prediction accuracy of fraud detection systems and promoting innovation and improvement in the field of financial security.