Track: Machine Learning

Abstract

The notion of risk exists in every aspect of the business, as it cannot be eliminated but rather reduced to an attainable level through the utilization of effective risk management techniques. For the insurance industry in particular, risk is traded and transferred to the insurance providers as the company offers a shield from the exposure to risk consequences and the likelihood of loss, therefore, escalating the risk from the insured entity to the insurer for a given premium. This research is a development on a previously published paper by the author which had focused on the same issue but with the utilization of binary regression. The paper now proposes a modern model to risk classification which will be used for property lines insurance. The significance of the research lies in the fact that the process of risk prediction can be extremely complex since there are many parameters to keep in mind.  In addition, accurate premium pricing can be difficult to estimate given the unpredictability of the risks occurring to the covered property. The data will be categorized into “A”, “B”, “C”, and “D” using multinomial regression followed by a pricing model. The proposed model will be validated via data collected from surveyed properties of a UAE based insurance company. The model is expected to serve as a tool that helps provide better estimates of risk, premiums, and precise pricing.