Several collective risk models have been recently published that relaxed the unrealistic assumption of independence between claim frequency and severities. However, most of these models considered the dependence between the frequency and average/aggregated severities. Now, Oh et al. (2020) propose an alternative model to capture the dependence between the frequency and the individual severities, using elliptical copulas, as well as generalizing it to vine copulas. Copula models over a exible way to model the dependence structure along with the marginal distributions. However, in the case of copula models with discrete and continuous marginals, the density function cannot be conveniently separated as with models with only continuous margins. Since collective risk models, like the ones Oh et al. (2020), have discrete marginal distributions for the count random variable and continuous distributions for the severities random variables, we apply a copula transformation method, introduced in Ahn et al. (2020) to it, to obtain a model that behaves as if all marginals are continuous. This makes the interpretation of the model more intuitive and its application broader. Along with applying this transformation method to the vine copula model from Oh et al. (2020) in this thesis, two distribution possibilities are considered for the count random variable and several estimation algorithms are tested for their performance for the transformed collective risk model. Following this analysis, a further illustration of the model is presented by analyzing an automobile dataset.     «

Several collective risk models have been recently published that relaxed the unrealistic assumption of independence between claim frequency and severities. However, most of these models considered the dependence between the frequency and average/aggregated severities. Now, Oh et al. (2020) propose an alternative model to capture the dependence between the frequency and the individual severities, using elliptical copulas, as well as generalizing it to vine copulas. Copula models over a exible way...     »