Driver risk classification in auto insurance: Using deep neural networks and in-vehicle CAN bus data
Date
2020
item.page.datecreated
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Among the numerous ways auto insurance companies can contribute to safety on the
roads whilst creating more value for themselves and customers is by effectively analyzing
individual driver risk level. Every driver’s driving style is their fingerprint and when
insurance companies can anticipate their style and the risk associated with it, they can create
safer portfolios, reward good driving with lower premiums, customize customer offerings
and penalize bad driving with higher premiums. A good driver risk assessment lies in
identifying and analyzing behavioral patterns in driving. The challenge however with
prevalent risk assessment methods in motor insurance are that they rely on non-precise data
(age, occupation, address) and their assessment is merely descriptive, leaving little to no
detail about the individual nature of the risk a driver might pose. In this project I design a
web interface for users and use In-vehicle sensor data that identify driver behaviour patterns
and a deep neural network with continuous learning capabilities to analyse and predict the
driver’s risk based on the data from their vehicle.
Description
Applied project submitted to the Department of Computer Science and Information Systems, Ashesi University, in partial fulfillment of Bachelor of Science degree in Management Information Systems, May 2020
item.page.type
Applied project
item.page.format
Keywords
auto insurance, driver risk analysis system, web-based application