Improving Road Safety with Machine Learning



Piers Campbell


Data Science Lead - Kainos

Abstract: The DVSA’s primary objective is to make GB’s roads safer. The original approach to risk rating MOT testing garages is subjective and open to dispute, focused on garage performance only and does not fully leverage the data available in the MOT Testing Service (MTS).

DVSA and Kainos have pioneered a predictive analytics tool using Machine Learning (ML) that derives risk scores for garages and testers based on data collected and compared against other testers and garages in the scheme. These scores are combined with knowledge of current and past disciplinary action to derive an overall risk rating.

Using this tool, DVSA can now more effectively and efficiently target their inspection and enforcement action at garages and testers exhibiting unusual testing patterns, therefore making a greater positive impact on road safety.

The project is an excellent example of private and public-sector collaboration and innovation, which highlights the benefits of leveraging Machine Learning techniques to:

• Better utilise operational data for decision making in a public service;

• Improve efficiency and effectiveness of an enforcement team with modest resources;

• Use predictive analytics to meet the strategic objectives of a government agency

Crucially, using the data-based evidence from the risk rating tool is reducing the number of disputed disciplinary decisions which means that the enforcement staff can be considerably more effective and efficient in what they do to achieve the strategic target of increasing disciplinary cases by 8%.

The impact of this collaborative project was recognised when the it was selected as the inaugural UK IT Awards AI/AL project of the year in November 2018.

Bio: Dr Piers Campbell leads data science within the Data and Analytics capability at Kainos. Piers has significant experience in data driven decision making and predictive models across multiple domains. Piers has designed and delivered advanced analytics solutions for Government, Telecos, Financial Services and Professional Sports Teams. Prior to joining Kainos Piers worked with both Accenture and Deloitte having previously spent eight years in academic roles at universities in the UK, Middle East and Australia.