A Machine Learning Based Model For Prevention of Fraud in E-commerce Transactions

Abstract

In this study I worked on the prevention of fraud in online transactions and proposed a Machine Learning based multilayer model and implemented it on a real business selling digital products online. The results as shown in chapter 4 suggest that the proposed solution is highly effective in fraud prevention and is able to reduce the fraudulent transaction by more than 50%. Proposed model is also highly adaptive to dynamic business needs and can be configured to tighten or loosen the fraud control strategy of an online business. This work can further be extended by adding spike and communal detection techniques in the model. A layer for the assessment of economic efficiency of model can also be made part of this model, which can be helpful in reducing the number of refunds a business has to make.