arun-gharami/lead-ai-fraud-detection-dataset-v2

Lead.AI欺诈检测数据集v2（研究级）是一个合成的、研究级的表格数据集，用于欺诈检测。它包含100,000条合成交易记录，每条记录标记为risk_label = 1（欺诈）或risk_label = 0（合法）。数据集包含20个特征，包括交易详情、客户行为和上下文信息，特别设计用于支持可解释AI（XAI）研究。数据集通过复杂的特征间相关性和明确的信号，提供了一个具有挑战性但结构化的环境，用于开发和评估高级欺诈检测模型及其可解释性。数据集还考虑了偏差和公平性问题，并提供了详细的数据生成方法和使用指南。

Lead.AI Fraud Detection Dataset v2 (Research-Grade) is a synthetic, research-grade tabular dataset for fraud detection. It contains 100,000 synthetic transaction records, each labeled as either risk_label = 1 (fraudulent) or risk_label = 0 (legitimate). The dataset includes 20 features covering transaction details, customer behavior and contextual information, and is specifically designed to support explainable AI (XAI) research. It provides a challenging yet structured environment for developing and evaluating advanced fraud detection models and their interpretability through complex inter-feature correlations and explicit signals. The dataset also considers bias and fairness issues, and provides detailed data generation methods and usage guidelines.