Over the last ten years, there has been an explosion in the growth of powerful machine learning and predictive analytics techniques. Spurred on by the surge of data from the internet and the availability of relatively cheap and powerful computing, data scientists and engineers have developed extremely powerful methods and algorithms for extracting information from experience records and predicting future events.
Elucidor brings these advanced techniques to the insurance industry. Our modeling engine, DeepInsight!, incorporates generalized linear modeling, Bayesian inference and survival analysis methodologies.
DeepInsight! is a powerful risk analytics engine which determines to what degree “predictor” variables will affect an insured risk. The system offers significant advantages over traditional actuarial approaches to inferring risk parameters. These advantages include:
- The analysis can uncover relationships between predictor and predicted variables with far less data than using traditional techniques.
- Complex relationships between predictor variables can be analyzed. For example, when analyzing impaired life mortality, the effect of cardiac impairment and diabetes could be larger than the impact of the sum of the two individual effects (this is the co-morbidity effect).
- Since a Bayesian approach is adopted, the analysis allows for the incorporation of “prior” or existing assumptions such as the existing mortality, morbidity tables or the incidence and severity rates for P&C or health business. DeepInsight! analytical results are then provided on a “relative” basis, and show by what proportion the “prior” or existing assumptions should be adjusted given the emerging data.