1. Proceedings
  2. I3M
  3. 2022
  4. EMSS
  5. 10.46354/i3m.2022.emss.039

Customer Behaviour Classification Using Simulated Transactional Data 

  • a Ryan Butler ,
  • b Ethan Hinton, 
  • c Max Kirwan, 
  • d Abraham Salih
  • a,b,c,d Department of Engineering Mathematics, University of Bristol, Bristol, BS8 1TR, UK 
Cite as
Butler R., Hinton E., Kirwan M., and Salih A. (2022).,Customer Behaviour Classification Using Simulated Transactional Data. Proceedings of the 34th European Modeling & Simulation Symposium (EMSS 2022). , 039 . DOI: https://doi.org/10.46354/i3m.2022.emss.039

Abstract

The practice of commercial banks using transactional data to improve customer experience and profitability has become widespread, due to both the progression of a cashless society and advances in machine learning and data science, which has enabled the large-scale processing of such data. Current research has focused on reducing the risk of credit events by analysing transactional data using various machine learning techniques and, more recently, neural networks. However, the use of these techniques for the marketing of retail bank products is limited in the current literature. This paper introduces a novel technique to combine data of both where and when customers spend money, utilising a Convolutional Neural Network (CNN) classifier, trained on a visual representation of a customer’s transaction history. From this, customer spending behavior in each transaction category can be labelled. Distinct clusters of accounts were also identified based on their spending habits and personality type, using two simulated transactional datasets produced via agent-based modelling and provided to us by Lloyds Banking Group, a major UK retail bank. All data supplied by Lloyds Banking Group was computer generated, synthetic data. At no time was real transactional data shared. The techniques introduced in this paper could be used by commercial banks to improve marketing strategies.

References

  1. Albawi, S., Mohammed, T. A., and Al-Zawi, S. (2017). Un derstanding of a convolutional neural network. 2017 International Conference on Engineering and Technology (ICET), pages 1–6.
  2. Best, R. d. (2022). Card payments per day uk. https://www.statista.com/statistics/719708/ card-payments-per-day-forecast-united-kingdom/. 
  3. Brownlee, J. (2020). A gentle introduction to k-fold cross-validation. https://machinelearningmastery. com/k-fold-cross-validation. 
  4. Campello, R. J., Moulavi, D., and Sander, J. (2013). Density based clustering based on hierarchical density esti mates. Advances in Knowledge Discovery and DataMining, page 160–172. 
  5. Carter, C. and Catlett, J. (1987). Assessing credit card appli cations using machine learning. IEEE Expert, 2:71–79. Gandhi, R. (2018). Build your own convolution neural network in 5 mins. 
  6. Giannotti, F., Gozzi, C., and Manco, G. (2002). Clustering transactional data. In European Conference on Principles of Data Mining and Knowledge Discovery, pages 175–187. Springer. 
  7. Hochreiter, S. and Schmidhuber, J. (1997). Long short term memory. Neural Computation, 9(8):1735–1780. Holm, M. (2018). Machine learning and spending patterns : 
  8. A study on the possibility of identifying riskily spending behaviour. Master’s thesis, KTH, School of Computer Science and Communication (CSC). 
  9. Khandani, A. E., Kim, A. J., and Lo, A. (2010). Consumer credit-risk models via machine-learning algorithms. Journal of Banking Finance, 34(11):2767–2787. 
  10. Koehler, M., Tivnan, B., and Bloedorn, E. (2005). Gener ating Fraud: Agent Based Financial Network Modeling. page 5. 
  11. LeCun, Y. and Bengio, Y. (1995). Convolutional networks for images, speech, and time-series. MIT Press. 
  12. LeCun, Y., Bengio, Y., and Hinton, G. E. (2015). Deep learn ing. Nat., 521(7553):436–444.