title: "Rgadget" output: html_document: keep_md: yes
Rgadget is a set of useful utilities for gadget, a statistical multi-species multi-area marine ecosystem modelling toolbox.
This package aids in the developement of Gadget models in a number of ways. It can interact with Gadget, by manipulating input files, digest output and rudimentary plots. A rendered version of the on-line help can be accessed here
Installing
You can use devtools to install this directly:
# install.packages("devtools")
devtools::install_github("gadget-framework/rgadget")Using
To use Rgadget you will need to load it into memory:
library(Rgadget)
theme_set(theme_light()) ## set the plot theme (optional)
library(patchwork) ## optional packages
scale_fill_crayola <- function(n = 100, ...) {
# taken from RColorBrewer::brewer.pal(12, "Paired")
pal <- c("#A6CEE3", "#1F78B4", "#B2DF8A", "#33A02C",
"#FB9A99", "#E31A1C", "#FDBF6F", "#FF7F00",
"#CAB2D6", "#6A3D9A", "#FFFF99", "#B15928")
pal <- rep(pal, n)
ggplot2::scale_fill_manual(values = pal, ...)
}To illustrate the use of Gadget we will use a model for cod in Icelandic waters atteched to the package. You can access the model using the following code:
system.file('extdata', 'cod_model.tgz', package = 'Rgadget') %>%
untar(exdir = path.expand('./gadget_example/'))
## change the working directory to the location of the gadget model
setwd('gadget_example/cod_model')To estimate the model parameters the suggested procedure is to use the iterative reweighting approach with is implemented in the gadget.iterative function (see ?gadget.iterative for further details).
gadget.iterative(main='main',
grouping=list(sind1=c('si.gp1','si.gp1a'),
sind2=c('si.gp2','si.gp2a'),
sind3=c('si.gp3','si.gp3a')),
params.file = 'params.in',
wgts='WGTS')This function calls Gadget which behind the scenes does the parameter estimation which we will use. To obtain information on the model fit and properties of the model one can use the gadget.fit function to query the model:
fit <- gadget.fit()The fit object is essentially a list of data.frames that contain the likelihood data merged with the model output.
fit %>% names()## [1] "sidat" "resTable" "nesTable"
## [4] "suitability" "stock.growth" "stock.recruitment"
## [7] "res.by.year" "stomachcontent" "likelihoodsummary"
## [10] "catchdist.fleets" "stockdist" "SS"
## [13] "stock.full" "stock.std" "stock.prey"
## [16] "fleet.info" "predator.prey" "params"
## [19] "catchstatistics"and one can access those data.frames simply by calling their name:
fit$sidat## # A tibble: 135 x 20
## name area label year step number intercept slope sse stocknames
## <chr> <chr> <chr> <int> <int> <dbl> <dbl> <dbl> <dbl> <chr>
## 1 si.g… area1 leng… 1985 2 2.31e8 -35.3 2.21 4.53 codimm
## 2 si.g… area1 leng… 1986 2 1.53e8 -35.3 2.21 4.53 codimm
## 3 si.g… area1 leng… 1987 2 8.71e7 -35.3 2.21 4.53 codimm
## 4 si.g… area1 leng… 1988 2 1.18e8 -35.3 2.21 4.53 codimm
## 5 si.g… area1 leng… 1989 2 9.48e7 -35.3 2.21 4.53 codimm
## 6 si.g… area1 leng… 1990 2 1.48e8 -35.3 2.21 4.53 codimm
## 7 si.g… area1 leng… 1991 2 1.13e8 -35.3 2.21 4.53 codimm
## 8 si.g… area1 leng… 1992 2 5.16e7 -35.3 2.21 4.53 codimm
## 9 si.g… area1 leng… 1993 2 1.20e8 -35.3 2.21 4.53 codimm
## 10 si.g… area1 leng… 1994 2 1.58e8 -35.3 2.21 4.53 codimm
## # … with 125 more rows, and 10 more variables: sitype <chr>,
## # fittype <chr>, length <chr>, age <chr>, survey <chr>, fleet <chr>,
## # observed <int>, lower <int>, upper <int>, predict <dbl>For further information on what the relevant data.frames contain refer to the help page for gadget.fit.
In addition a plot routine for the fit object is implement in Rgadget. The input to the plot function is simply the gadget.fit object, the data set one wants to plot and the type. The default plot is a survey index plot:
plot(fit)
To produce a likelihood summary:
plot(fit,data='summary')
A weighted summary plot:
plot(fit,data='summary',type = 'weighted')
and an pie chart of likelihood components:
plot(fit,data='summary',type='pie')## Warning in RColorBrewer::brewer.pal(n, pal): n too large, allowed maximum for palette Spectral is 11
## Returning the palette you asked for with that many colors
To plot the fit to catch proportions (either length or age) you simply do:
tmp <- plot(fit,data = 'catchdist.fleets')
names(tmp)## [1] "alkeys.aut" "alkeys.comm" "alkeys.igfs" "ldist.aut" "ldist.comm"
## [6] "ldist.igfs"and then plot them one by one:
tmp$alkeys.aut
tmp$ldist.aut
One can also produce bubble plots
bubbles <- plot(fit,data = 'catchdist.fleets',type='bubble')## Joining, by = "name"
## Joining, by = "name"names(bubbles)## [1] "ldist" "aldist"Age bubbles
bubbles$aldist
Length bubbles
bubbles$ldist
One can also illustrate the fit to growth in the model:
grplot <- plot(fit,data = 'catchdist.fleets',type='growth')
names(grplot)## [1] "alkeys.aut" "alkeys.comm" "alkeys.igfs"Illstrate the fit to the autumn survey
grplot$alkeys.aut## Warning: Removed 1 rows containing missing values (geom_point).## Warning: Removed 4 rows containing missing values (geom_linerange).
And the fit to maturity data:
plot(fit,data='stockdist')## $matp.igfs## Warning: Removed 52 rows containing missing values (geom_point).
And selection by year and step
plot(fit,data="suitability")
Age age compostion
plot(fit,data='stock.std') + scale_fill_crayola()
And the standard ices plots
plot(fit,data='res.by.year',type='total') + theme(legend.position = 'none') +
plot(fit,data='res.by.year',type='F') + theme(legend.position = 'none') +
plot(fit,data = 'res.by.year',type='catch') + theme(legend.position = 'none') +
plot(fit, data='res.by.year',type='rec')## Warning: Removed 52 rows containing missing values (geom_path).
Acknowledgements
This project has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no.613571.