A Prioritized Fraud Detection Model for Subscription-Based Businesses Platform with a Minimal Labelled Data
In this era of e-commerce, many companies are moving towards the subscription-based online business model to deliver services directly to consumers. Fraudsters are using these online platforms for different types of malicious activities. Most often, identifying fraudsters is challenging for many companies due to the limitation of time and other resources. On the other hand, a fully automated fraud detection model (FDM) also creates a risk of true-negative identification. In this paper, we develop a Prioritized Fraud Detection Model (PFDM) that generates a prioritized list of fraud risk accounts which use a minimum labelled dataset. This model consists of a trained unsupervised clustering and a neural network classifier. We use this model to identify fraud accounts in online subscription-based business data. Our model shows promising in terms of practical use cases where each suspicious fraud case requires to examine by a human.