A mature UK Motor insurer worked with HAL to rebuild key existing pricing models using Gradient Boosting Machines (GBMs). Insurers increasingly use machine learning models like GBMs to automate and increase the accuracy of their pricing models. However, GBMs are more challenging to interpret and communicate than conventional modeling approaches.
Our experts used our proprietary model to build a custom and interactive GBM visualisation solution to review GBM performance and explain the impact of individual data features on the model output. This made model review and sign-off faster and more insightful as the complexity of the GBM was transformed into an interactive and easy-to-understand output.