Gamma Mixture Density Networks and their application to modelling insurance claim amounts

Abstract We discuss how mixtures of Gamma distributions with mixing probabilities, shape and rate parameters depending on features can be fitted with neural networks. We develop two versions of the EM algorithm for fitting so-called Gamma Mixture Density Networks, which we call the EM network boosting algorithm and the EM forward network algorithm, and we test their implementation together with the choices of hyperparameters. A simulation study shows that our algorithms perform very well on synthetic data sets. We further illustrate the application of the Gamma Mixture Density Network on a real data set of motor insurance claim amounts and conclude that Gamma Mixture Density Networks can improve the fit of the regression model and the predictions of the claim severities used for rate-making compared to classical actuarial techniques.

[1]  Michael I. Jordan,et al.  Hierarchical Mixtures of Experts and the EM Algorithm , 1994, Neural Computation.

[2]  Timothy Dozat,et al.  Incorporating Nesterov Momentum into Adam , 2016 .

[3]  A CLASS OF MIXTURE OF EXPERTS MODELS FOR GENERAL INSURANCE: APPLICATION TO CORRELATED CLAIM FREQUENCIES , 2019, ASTIN Bulletin.

[4]  Tsz Chai Fung,et al.  A class of mixture of experts models for general insurance: Theoretical developments , 2019 .

[5]  Christopher N Davis,et al.  The use of mixture density networks in the emulation of complex epidemiological individual-based models , 2019, bioRxiv.

[6]  Paul D. McNicholas,et al.  Modeling frequency and severity of claims with the zero-inflated generalized cluster-weighted models , 2020 .

[7]  F. Leisch,et al.  Finite Mixtures of Generalized Linear Regression Models , 2008 .

[8]  Geoffrey E. Hinton,et al.  Adaptive Mixtures of Local Experts , 1991, Neural Computation.

[9]  Pietro Parodi.  A GENERALISED PROPERTY EXPOSURE RATING FRAMEWORK THAT INCORPORATES SCALE-INDEPENDENT LOSSES AND MAXIMUM POSSIBLE LOSS UNCERTAINTY , 2020 .

[10]  Bettina Grün,et al.  Modeling loss data using mixtures of distributions , 2016 .

[11]  X. Sheldon Lin,et al.  Modeling and Evaluating Insurance Losses Via Mixtures of Erlang Distributions , 2010 .

[12]  Martin Blostein,et al.  On modeling left-truncated loss data using mixtures of distributions , 2019, Insurance: Mathematics and Economics.

[13]  B. Jørgensen.  Exponential Dispersion Models , 1987 .

[14]  Yoshua Bengio,et al.  Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[15]  Peter K. Dunn,et al.  Randomized Quantile Residuals , 1996 .

[16]  Giovanni Parmigiani,et al.  GAMMA SHAPE MIXTURES FOR HEAVY-TAILED DISTRIBUTIONS , 2008, 0807.4663.

[17]  W. DeSarbo,et al.  A mixture likelihood approach for generalized linear models , 1995 .

[18]  Daniel Fernández,et al.  On Two Mixture-Based Clustering Approaches Used in Modeling an Insurance Portfolio , 2018 .

[19]  Xi Chen,et al.  Finite mixture-of-gamma distributions: estimation, inference, and model-based clustering , 2019, Advances in Data Analysis and Classification.

[20]  Dimitris Karlis,et al.  A finite mixture of bivariate Poisson regression models with an application to insurance ratemaking , 2012, Comput. Stat. Data Anal..

[21]  D. Karlis,et al.  AN EM ALGORITHM FOR FITTING A NEW CLASS OF MIXED EXPONENTIAL REGRESSION MODELS WITH VARYING DISPERSION , 2020 .

[22]  Roel Verbelen,et al.  FITTING MIXTURES OF ERLANGS TO CENSORED AND TRUNCATED DATA USING THE EM ALGORITHM , 2014, ASTIN Bulletin.

[23]  X. Sheldon Lin,et al.  Efficient Estimation of Erlang Mixtures Using iSCAD Penalty with Insurance Application , 2016 .

[24]  Wenyong Gui,et al.  Fitting the Erlang mixture model to data via a GEM-CMM algorithm , 2018, J. Comput. Appl. Math..