Modelibrary.

05 Automated advisory

Automated advisory
in car insurance.

Client
Insurance company
Line
Car insurance
Method
Unsupervised learning

Risk rankings

The project helped an insurance company automate aspects of expert assessment when ranking cars and policyholders by risk.

Policyholder grouping

Using unsupervised learning, we created a three-group ranking of policyholders.

The ranking was partly consistent with previously known figures.

Fig. 01

Three groups, each drawn as a stepped profile of bars, set against one dashed line marking the common average.
Three policyholder groups shown against a common average.Illustration of the method, not client data.

Fig. 02

Four clusters of points, each with its centre marked by a cross, separated by dashed boundary lines.
Clusters with their centroids and the boundaries between them.Illustration of the method, not client data.

Pricing a similar risk? Get in touch

Hero artwork is original abstract brand imagery. It is not Modelibrary data, a client result, a calibrated probability or a performance claim. This site uses no cookies or analytics. Contact us by email.