Machine Learning in Insurance

• Other Authors: Nielsen, Jens Perch (Editor), Asimit, Alexandru (Editor), Kyriakou, Ioannis (Editor)
• Format: Electronic Book Chapter
• Language: English
• Published: Basel, Switzerland MDPI - Multidisciplinary Digital Publishing Institute 2020
• Subjects: History of engineering & technology; deposit insurance; implied volatility; static arbitrage; parameterization; machine learning; calibration; dichotomous response; predictive model; tree boosting; GLM; validation; generalised linear modelling; zero-inflated poisson model; telematics; benchmark; cross-validation; prediction; stock return volatility; long-term forecasts; overlapping returns; autocorrelation; chain ladder; Bornhuetter-Ferguson; maximum likelihood; exponential families; canonical parameters; prior knowledge; accelerated failure time model; chain-ladder method; local linear kernel estimation; non-life reserving; operational time; zero-inflation; overdispersion; automobile insurance; risk classification; risk selection; least-squares monte carlo method; proxy modeling; life insurance; Solvency II; claims prediction; export credit insurance; semiparametric modeling; VaR estimation; analyzing financial data; n/a
• Online Access: DOAB: download the publication; DOAB: description of the publication

Machine learning is a relatively new field, without a unanimous definition. In many ways, actuaries have been machine learners. In both pricing and reserving, but also more recently in capital modelling, actuaries have combined statistical methodology with a deep understanding of the problem at hand and how any solution may affect the company and its customers. One aspect that has, perhaps, not been so well developed among actuaries is validation. Discussions among actuaries' "preferred methods" were often without solid scientific arguments, including validation of the case at hand. Through this collection, we aim to promote a good practice of machine learning in insurance, considering the following three key issues: a) who is the client, or sponsor, or otherwise interested real-life target of the study? b) The reason for working with a particular data set and a clarification of the available extra knowledge, that we also call prior knowledge, besides the data set alone. c) A mathematical statistical argument for the validation procedure.