Proxy modelling using machine learning: WCA 2020 lecture

This years Worshipful Company of Actuaries lecture is delivered by Guarang Mehta, FIA

Gaurang Mehta

Wed 19 February 2020, Sibson Lecture Theatre 3, 13:30

Each year the School is pleased to host a guest lecture sponsored by the Worshipful Company of Actuaries*. This year’s lecture looks at advances in proxy modelling techniques (abstract below) and is delivered by Guarang Mehta, an alumnus of the School.

The lecture is on Wednesday 19 February 2020 in Sibson Lecture Theatre 3, beginning at 13:30. There will be refreshments and networking opportunities afterwards in the Sibson atrium.

To attend the event, which is free of charge, please register by sending a blank email to wcalecture@kent.ac.uk with the subject “Register”.

Abstract for the lecture:

Use of proxy modelling is prevalent in the insurance sector under Solvency II and so far efforts have been made to solve this problem as a least square regression problem. We have explored solving the same problem using state-of-art machine learning techniques and also improving the same least square regression methods with aid from machine learning techniques such as feature importance and cross validation. We show that material improvement in the fits can be achieved by combining machine learning techniques and provide a few recommendations for practitioners working in the area of proxy models.

*  Established in 1979, the Worshipful Company of Actuaries is a sociable and philanthropic community of actuaries and a Livery Company of the City of London.