electricsheepafrica/nigerian-banking-agent-network

这是一个基于尼日利亚银行代理及其交易的生产级合成数据集，包含真实的尼日利亚银行环境，旨在用于代理欺诈检测、网络优化和佣金管理。数据集具有尼日利亚的银行名称、各州和人口统计信息，机器学习就绪的标签（欺诈标志，欺诈发生率为2.5%），覆盖2023年至2024年的时序数据，共200万行数据，分为训练集、验证集和测试集，关键列包括代理ID、名称、州、类型（超级/主/子）、月交易额、月交易数、佣金率、身份验证级别和欺诈标志。

This is a production-grade synthetic dataset with authentic Nigerian banking context, designed for agent fraud detection, network optimization, and commission management. The dataset includes real bank names, states, demographics of Nigeria, machine learning-ready labels with a fraud flag at a 2.5% prevalence, temporal coverage from 2023 to 2024, with a total of 2,000,000 rows split into training, validation, and test sets, with key columns such as agent_id, agent_name, state, agent_type (super/master/sub), monthly_volume_ngn, monthly_transactions, commission_rate, kyc_tier, and fraud_flag.