Synthetic Mobile Money Transaction Dataset

Published: 29 October 2024| Version 2 | DOI: 10.17632/zhj366m53p.2
Contributor:
Denish Azamuke

Description

This dataset comprises synthetic mobile money transaction records, meticulously generated to closely resemble real transaction data for training machine learning models in financial fraud detection. It was produced using the MoMTSim platform, a multi-agent-based simulation specifically tailored to emulate the mobile money financial ecosystem. The simulation inputs are based on real financial data to ensure realism in transaction dynamics. The dataset's fidelity to real-world data has been validated through several statistical methods including the sum of squared errors approach, the Kolmogorov-Smirnov test, and Bland-Altman plots, confirming its accuracy and utility for research and practical applications in detecting fraudulent transactions.

Files

Institutions

Makerere University

Categories

Artificial Intelligence, Data Science, Machine Learning

Funding

JPMorgan Chase & Co (United States)

Digital Credit Observatory (DCO), a program of the Center for Effective Global Action (CEGA), with support from the Bill & Melinda Gates Foundation

Google PhD Fellowship Program

Licence