Get Retrospective Plot
get_retrospective_plot.RdGenerates three retrospective diagnostic plots from a set of peeled model runs: a relative-difference plot with Mohn's rho, an absolute-scale trajectory plot, and a cohort-tracking squid plot for recruitment. Together these plots diagnose systematic retrospective patterns in SSB and recruitment estimates across sequential data removals.
Arguments
- retro_output
Data frame produced by
do_retrospective, containing columnsYear,value,peel,Type(e.g."SSB","Recruitment"),Pop, andRegion. Rows withpeel = 0represent the full-data terminal run; rows withpeel > 0represent sequential data removals. Rows withvalue = 0are excluded before plotting.- Rec_Age
Integer giving the age at recruitment (e.g.
2for age-2 recruitment). Used to convert model year to cohort year in the squid plot (cohort = Year - Rec_Age).
Value
A list of three ggplot objects:
- [[1]] retro_plot
Relative-difference plot. Each line shows the proportional deviation of a peeled run from the terminal-year estimate at each year, computed via
get_retrospective_relative_difference. Terminal-year points (wherepeel = max(Year) - Year) are highlighted. Mohn's rho (mean relative difference across terminal-year points) is annotated on each facet panel. Faceted by Region × (Type + Population); lines and points coloured by peel year on a continuous viridis scale.- [[2]] abs_retro_plot
Absolute-scale trajectory plot. Peeled runs (
peel > 0) are drawn as solid lines coloured by peel; the full-data run (peel = 0) is overlaid as a dashed black line. Faceted by Region × (Type + Population) with free y-scales.- [[3]] squid_plot
Cohort-tracking (squid) plot for recruitment. Restricted to the 10 most recent cohorts. X-axis shows years since the cohort was last estimated (
terminal - Year - 1), y-axis shows recruitment value, and lines are grouped and coloured by cohort year. Faceted by Population × Region with free y-scales.
Examples
if (FALSE) { # \dontrun{
retro <- do_retrospective(
n_retro = 7,
data = data,
parameters = parameters,
mapping = mapping,
random = NULL,
do_par = TRUE,
n_cores = 7,
do_francis = FALSE,
n_francis_iter = NULL
)
plots <- get_retrospective_plot(retro, Rec_Age = 2)
plots[[1]] # relative difference + Mohn's rho
plots[[2]] # absolute trajectories
plots[[3]] # squid plot
} # }