MetricsWeighted: deviance_gamma – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

deviance_gamma

Gamma Deviance

Description

Weighted average of (unscaled) unit Gamma deviance, see e.g. the reference below. Special case of Tweedie deviance with Tweedie parameter 2. The smaller the deviance, the better.

Usage

deviance_gamma(actual, predicted, w = NULL, ...)

Arguments

`actual`	Strictly positive observed values.
`predicted`	Strictly positive predicted values.
`w`	Optional case weights.
`...`	Further arguments passed to `weighted_mean`.

Value

A numeric vector of length one.

References

Jorgensen, B. (1997). The Theory of Dispersion Models. Chapman & Hall/CRC. ISBN 978-0412997112.

Examples

deviance_gamma(1:10, c(1:9, 12))
deviance_gamma(1:10, c(1:9, 12), w = rep(1, 10))
deviance_tweedie(1:10, c(1:9, 12), tweedie_p = 2)
deviance_tweedie(1:10, c(1:9, 12), tweedie_p = 1.99)
deviance_gamma(1:10, c(1:9, 12), w = 1:10)

MetricsWeighted

Weighted Metrics, Scoring Functions and Performance Measures for Machine Learning

v0.5.2

GPL (>= 2)

Authors

Michael Mayer [aut, cre, cph], Christian Lorentzen [ctb, rev]

Initial release

2021-04-16