Skip to contents

The weight of the experience rating against the exposure rating, by limited fluctuation credibility: \(Z = \min(1, \sqrt{n / n_{full}})\) for \(n\) losses, with \(n_{full}\) the number of losses for full credibility. The blended rate is \(Z\) times the experience rate plus \(1 - Z\) times the exposure rate.

Usage

credibility_weight(n, full = 1082)

Arguments

n

The number of losses in the layer (or above the threshold).

full

The number of losses for full credibility; the default, 1082, gives the observed frequency within 5% of the true one with a probability of 90% for Poisson counts.

Value

The credibility weight, between 0 and 1.

Examples

credibility_weight(25)
#> [1] 0.1520045
credibility_weight(25, full = 100)
#> [1] 0.5