Add delta-AIC and Akaike weights to a comparison table

Adds deltaAIC and AICweight columns to a data frame that carries an AIC column, and sorts the result so the best-supported model comes first.

Usage

add_aic_weights(x, group = NULL, aic_col = "AIC")

Arguments

x

A data frame with an AIC column.

group

Optional column name whose levels delimit comparable sets.

aic_col

Name of the AIC column. Default "AIC".

Value

x with deltaAIC and AICweight columns, sorted by AIC within group.

Details

Akaike weights are only meaningful among models fitted to the same response data. A table that mixes fits on different subsets — a Tweedie over all hauls beside a Gamma over the positive ones, say, or a biomass model beside an abundance model — has no common likelihood scale, and weights computed across it would invite exactly the comparison the table should be warning against. group exists for that case: weights are computed within each group, and left NA for groups with a single member.

See also

sdmTMB_selection()

Author

Mikko Vihtakari

Examples

tab <- data.frame(
  model = c("a", "b", "c"),
  AIC = c(1002, 1000, 1010)
)
add_aic_weights(tab)
#> # A tibble: 3 × 4
#>   model   AIC deltaAIC AICweight
#>   <chr> <dbl>    <dbl>     <dbl>
#> 1 b      1000        0   0.727  
#> 2 a      1002        2   0.268  
#> 3 c      1010       10   0.00490