摘要：With the development of the machine learning (ML) technique and broad successful application, machine learning
is becoming more and more popular for data analytics in many industries. Insurance is no exception, and machine
learning techniques are used to build predictive models in Claims (Fraud, subrogation models), Marketing
(Segmentation, cross sell model, recommendation models), and Underwriting. However, for pricing models,
Generalized Linear Models (GLM) still dominates given its easy interpretation and well-established frame work.
Using a machine learning method to enhance the GLMs model is a challenge to the insurance industry especially
for actuarial modeling. This paper will discuss some potential ways to enhance the GLMs model with tree based
machine learning techniques and give a case study on territorial analysis, which would show significant
improvement on the predictive nature of the GLM model.
关键词：Machine learning; territorial analysis; generalized linear modeling;
