AvoidingConfusion

R and stan scripts to produce the outputs and figures of Spence et al. (2024) "Avoiding Confusion: Modelling Image Identification Surveys with Classification Errors", Methods in Ecology and Evolution.

ML functions.R contains the main functions required for the other scripts to run.
Poisson generating process (3.1).R runs the Poisson example in Section 3.1.
Poisson and NB generating processes (3.2).R runs the mixed Poisson - Negative Binomial example in Section 3.2.
Sensitivity_to_C (3.3).R contains the code for the comparison study in Section 3.3.
Zooplankton survey (4).R runs the analysis of the Zooplankton survey data in Section 4. This file also contains the code to produce Figures S7 and S8 in Section S5 of the Supporting Information.
Simulation-based calibration.R runs the simulation-based calibration in Supporting Information Section S3.
Zero-inflated Poisson with covariates.R runs the additional example with a covariate influencing the count generating process, see Section S6 of the Supporting Information.

poisson_model.stan contains the stan model used in Poisson generating process (3.1).R and Simulation-based calibration.R.
plankton_model.stan contains the stan model used in Zooplankton survey (4).R.
zeroinf_poisson_model.stan contains the stan model used in Zero-inflated Poisson with covariates.R.

plank_data.csv is the data used by Zooplankton survey (4).R.

Requirements

R version 4.3.2 or higher is required. The following R packages need to be installed to run the R scripts:

install.packages( c("rstan","ggplot2","reshape2",
                    "posterior","bayesplot"
                    "patchwork","MCMCprecision",
                    "VGAM","parallel", "MASS"))

Citation

Spence M., Barry J., Bartos T., Blackwell R., Scott J., Pitois S. 2024. “Avoiding Confusion: Modelling Image Identification Surveys with Classification Errors.” Methods in Ecology and Evolution. In press.