A method to determine model points with cluster analysis

A method to determine model points with cluster analysis


Thanks! Share it with your friends!


You disliked this video. Thanks for the feedback!

Sorry, only registred users can create playlists.


Speaker: Yosuke Goto

With the recent progresses such as the enforcement of Solvency II in Europe, the publication ICS version 1.0 and development of ERM, needs for timely calculation of indicators with huge computation is growing (e.g. risk measurements on a group entity or a profitability measurement on a stochastic method). While life insurance contracts in Japan and Europe generally have long duration and calculation of their cash flow usually takes a long time, calculation of cash flow in thousands of scenarios is required in stochastic methods. Therefore, an efficient method which reduces an amount of calculation is awaited.

This presentation shows a method that can determine model points mechanically by cluster analysis which is one of machine leaning technique. These model points will enable us to reduce an amount of calculation and to calculate indicators more flexibly. Furthermore, this method has an advantage that it does not require knowledge of a characteristic of insurance contracts because it determines model points automatically.
In this method, we classify cash flows of contracts into categories depending on similarity analyzed by cluster analysis, and choose a model contract from each category. We analyzed term insurance, endowment insurance, whole life insurance and single premium whole life insurance with dynamic lapse assumption by the method, and compared their accuracy.

Post your comment

Sign in or sign up to post comments.
Be the first to comment