Observed-vs-fitted residuals from an Rceattle fit — residuals.Rceattle • Rceattle

Returns a long-format data frame of residuals across one or more of the four fitted data sources: "index" (survey indices), "catch" (fishery catches), "comp" (age- or length-composition proportions), and "caal" (conditional age-at-length proportions).

Usage

# S3 method for class 'Rceattle'
residuals(object, type = "index", scale = "log", ...)

Arguments

object: An object of class "Rceattle" returned by fit_mod().
type: One or more of "index", "catch", "comp", "caal", or "all" (default "index").
scale: "log" (default) or "natural". Only affects "index" and "catch" residuals.
...: Currently unused.

Value

A data.frame with columns Source, Fleet_code, Fleet_name, Species, Sex, Year, Length, Bin, Age0_Length1, Sample_size, Observed, Fitted, Residual. Sex, Length, Bin, Age0_Length1, and Sample_size are NA where they do not apply (e.g. for index/catch rows).

Details

For "index" and "catch", the Residual column is on the log scale by default (matching the lognormal observation likelihood) and can be switched to the natural scale via scale = "natural". For "comp" and "caal", residuals are Pearson residuals on the fitted proportions: $$r = (p - \hat p)/\sqrt{\hat p (1 - \hat p)/N}$$ where N is the input sample size. Composition rows are returned in long form: one row per (observation, age/length bin).

Composition rows carry the Age0_Length1 flag from comp_data (0 for age comps, 1 for length comps) so age and length comps can be filtered apart. CAAL rows carry both the conditioning Length and the age Bin.