"Insurance claims caused by natural disasters exhibit spatial dependence with the strength of dependence being based on factors such as physical distance and population density, to name a few. Accounting for spatial dependence is therefore of crucial importance when modeling these types of claims. In this work, we present an approach to assess spatially dependent insurance risks using a combination of linear regression and factor copula models. Specifically, in loss modeling, observed dependence patterns are highly nonlinear, thus copula-based models seem appropriate since they can handle both linear and nonlinear dependence. The factor copula approach for estimating the spatial dependence reduces a complex dependence structure into a relatively easier task of estimating a spatial dependence parameter. Hence, we use a weighted sum of radial basis functions to model a spatial dependence parameter that determines the influence of each location. The methodology is illustrated using a thunderstorm wind loss dataset of Texas. Extensions to Matérn covariance functions and spatiotemporal models are briefly discussed"--Abstract, page iii.
Olbricht, Gayla R.
Samaranayake, V. A.
Mathematics and Statistics
M.S. in Applied Mathematics
Missouri University of Science and Technology
viii, 69 pages
© 2018 Tobias Merk, All rights reserved.
Thesis - Open Access
Electronic OCLC #
Merk, Tobias, "Models for high dimensional spatially correlated risks and application to thunderstorm loss data in Texas" (2018). Masters Theses. 7770.