dsims 0.2.0
----------
New Features
* Delta selection criteria is now recorded as the difference in informtion criteria between the top 2 best fitting models as determined by the information criteria.]
* The iteration numbers generating warnings or errors are now stored and displayed so user can choose what to do with these results.
Bug Fixes
* Fixed missing RMSE values
* Fix strata re-ordering for cluster size
* Models with -Inf information criteria no longer selected
* Models with dht = NULL are no longer selected
* Models which predict detection values < 1 no longer cause errors and are correctly excluded.
* Detectibility parameters for continuous covariates are now checked and validated.
* Fix situation where all reps are to be excluded due to problematic model fitting.
dsims 0.1.0
----------
Introducing the new Distance Sampling Simulation package!
dsims is our latest simulation package which interfaces with dssd so designs can be generated within R, thus making the simulation process a lot easier! Dsims also makes use of ggplot to produce cleaner looking graphics.
Region and Design
* dsims can make use of the region creation and all the designs currently in dssd.
Density
* dsims can generate density objects from constant values for each strata, from fitted mgcv gam objects with x and y as explantory covariates and from formulas of x and y.
* Density grids are stored as sf polygons with their associated x, y central coordinates and density value
Population Description
* Populations can either be created with fixed population sizes or based on the densities in the density grid.
* Both discrete and continuous individual level covariates can be included in the population
Detectablity
* The detectability of the population can be described by either half normal, hazard rate or uniform detection shapes. Parameters can vary by stratum
* Covariate parameters can be included to modify the scale parameter for each individual based on their covariate values.
Analyses
* A number of detection function analyses can be incorporated in a simulation and the model with the lowest criterion (AIC / AICc / BIC) will be selected.
* Defining analyses is based on the arguments which are passed to our Distance R library.
Simulations
* Simulations can be run in serial or parallel and their progress is output.
* The function run.survey can be used to create a single instance of a survey and check the simulation setup.