The Logit-weighted Reduced Mixture-of-Experts model is a flexible framework for modelling insurance claim numbers and claim amounts. Theoretical developments and some applications can be found in Fung et al. (2019) and Fung et al. (2019).
In this project, we implement this model in julia
as a statistical software package LRMoE.jl
. The source code and package documentation are given here and here. A paper accompanying this package has been published in the Annals of Actuarial Science (link).
We recommend using julia
as it provides great numerical performance. Meanwhile, an R
version is also available but is currently not regularly maintained (see here).
References:
Badescu, A.L., Fung, T.C., Lin, X.S. and Tseung, S.C. 2021. A better fit: presenting an intuitive and flexible non-linear regression model, The Actuary, pp.24-26. (Magazine article, not peer-reviewed)
Tseung, S.C., Badescu, A.L., Fung, T.C. and Lin, X.S., 2021. LRMoE. jl: a software package for insurance loss modelling using mixture of experts regression model. Annals of Actuarial Science, pp.1-22.
Tseung, S.C., Badescu, A., Fung, T.C. and Lin, X.S., 2020. LRMoE: an R package for flexible actuarial loss modelling using mixture of experts regression model. Available at SSRN 3740215.
Presentation: Actuarial Research Conference, Aug 2020