Install the development version from source:

# install.packages("devtools")
devtools::install_github("dragonfly-science/ComPoM")

Or from a local clone:

devtools::install()lation

You can install the development version from your local source using:

```r
# In the package root directory
devtools::install()

Or, if available on GitHub:

# Replace 'yourusername/ComPoM' with the actual repo
devtools::install_github("yourusername/ComPoM")

Capabilities

Function	Description
data_prep()	Prepares compositional count data for modelling
parse_form()	Constructs the Poisson-multinomial model formula
fit_model()	Fits the model using brms or sdmTMB backends
Fx_plot()	Plots compositional effects by group
scale_comps()	Scales compositions by a catch or effort variable
scaled_comp_plot()	Plots scaled compositions with uncertainty
scaled_ridge_plot()	Ridgeline plot of scaled compositions
post_pred_group()	Posterior predictive check by group

Basic Usage

library(ComPoM)

# 1. Prepare compositional count data
#    comp: data frame with columns for bin, count, and grouping variables
dat <- data_prep(
  comp             = my_data,
  vars_for_grouping = c("year", "area"),
  bin_lab          = "length_bin",
  count_var        = "count"
)

# 2. Fit the model (brms backend by default)
mod <- fit_model(
  form    = "year + area",
  data    = dat,
  backend = "brms",
  chains  = 4,
  iter    = 2000
)

# 3. Plot compositional effects
Fx_plot(mod, grp = "year")

# 4. Scale compositions by catch and plot
scaled <- scale_comps(
  scale_df = catch_data,
  predvar  = "catch",
  fit      = mod,
  grps     = c("year", "area")
)

scaled_comp_plot(scaled, grps = c("year", "area"))

Spatial model (sdmTMB backend)

library(sdmTMB)

mesh <- make_mesh(dat, xy_cols = c("x", "y"), cutoff = 20)

mod_spatial <- fit_model(
  form    = "year + area",
  data    = dat,
  backend = "TMB",
  mesh    = mesh,
  time    = "year"
)

Testing

Tests use testthat and run automatically during R CMD check.

devtools::test()