Claim count by mean and dispersion

The Panjer-class count with this mean and dispersion Var[N] / E[N]: binomial_count below 1, poisson_count at 1, negative_binomial_count above 1. A binomial needs a whole number of trials, so below 1 the trials are mean / (1 - dispersion) rounded up: the mean is kept and the dispersion moves up to the nearest attainable value.

Usage

claim_count(mean, dispersion)

Arguments

mean

Mean number of claims; finite and non-negative.

dispersion

Positive.

Value

A binomial_count, poisson_count or negative_binomial_count.

Examples

claim_count(4, 2.5)
#> <negative_binomial_count> r = 2.66666666666667, beta = 1.5
claim_count(10, 0.3)
#> <binomial_count> n = 15, p = 0.666666666666667