Title: OPTIMISING INSURANCE FRAUD DETECTION AND CLASSIFICATION OF VEHICLE ACCIDENT DAMANGE BY USING NEURAL NETWORKS TO IDENTIFY PATTERNS IN BEHAVIOUR, LINKED CASES AND VEHICLE RECOGNITION

PDF Images
Not Available
Year of Publication: 2013
Page Numbers: 611-620
Authors: Emma-Jane Phillips, Peter Phillips, Mark Hurrell
Conference Name: The Third International Conference on Digital Information Processing and Communications (ICDIPC2013)
- United Arab Emirates

Abstract:


This papers summarizes the use of neural networks as a business tool for analysis of insurance data in order to identify patterns. The primary focus of the paper examines the use of a neural network to classify vehicles are ‘writeoff’ or repairable in the UK insurance market, the expansion point of the work carried out was to expand the neural network use into the detection of fraud within the same market. Patterns are used to indicate fraud and also to determine the classification of an accident at the point of claim. The paper relates to a KTP project running through Northumbria university and indicates how a simple concept became a major business system that integrated technology with a business need to generate a social project that generated an income stream over 9 months before the project completion date.