Specify a claim frequency distribution

Creates a lightweight specification of a claim count distribution for use in actuarial risk models.

Usage

frequency_spec(dist, ..., custom = NULL, label = NULL)

Arguments

dist

Character scalar identifying the frequency distribution. Currently supported values are `"poisson"`, `"binomial"`, `"negative_binomial"`, `"geometric"`, and `"custom"`.

...

Named parameters defining the selected distribution.

custom

Optional named list defining a custom count distribution. Used only when `dist = "custom"`. At minimum, the list must contain `pmf`, a probability mass function.

label

Optional character scalar used to identify the specification.

Value

An object with classes `"tidyact_frequency_spec"` and `"tidyact_distribution_spec"`.

Details

A frequency specification represents a nonnegative integer-valued random variable. The constructor stores the probabilistic model only; it does not simulate observations, estimate parameters, or calculate aggregate losses.

Built-in parameterizations follow `stats`: Poisson uses `lambda`, binomial uses `size` and `prob`, negative binomial uses `size` and `prob`, and geometric uses `prob` with support beginning at zero.

Examples

frequency_spec("poisson", lambda = 4)
#> <tidyact_frequency_spec>
#> Distribution: Poisson
#> Parameters:
#>   lambda: 4
#> Support: {0, 1, 2, ...}
frequency_spec("binomial", size = 20, prob = 0.15)
#> <tidyact_frequency_spec>
#> Distribution: Binomial
#> Parameters:
#>   size: 20
#>   prob: 0.15
#> Support: {0, 1, ..., 20}
frequency_spec("negative_binomial", size = 2.5, prob = 0.4)
#> <tidyact_frequency_spec>
#> Distribution: Negative binomial
#> Parameters:
#>   size: 2.5
#>   prob: 0.4
#> Support: {0, 1, 2, ...}
frequency_spec("geometric", prob = 0.25)
#> <tidyact_frequency_spec>
#> Distribution: Geometric
#> Parameters:
#>   prob: 0.25
#> Support: {0, 1, 2, ...}