Development priorities

List of development priorities and rationale

Pending development priorities

Based on feedback from the community of users, as well as discussions with the steering committee, we will attempt to update a list of future development priorities here.

Change	Rationale
Explore oneStepPredict(.) for delta models	Allow better use of probability-integral-transform diagnostics
Add climate velocity as automated output based on raster of density predictions	Additional interpretation of existing (outputs)[https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.13295]
Add Hurlbert and other additional overlap metrics	Additional flexibility in (multivariate modelling)[https://onlinelibrary.wiley.com/doi/abs/10.1111/geb.12984]
Add stepwise AIC model-selection algorithm	Improve user interface when building models
Add options to automatically query publicly available data to population covariate_data to use in formula interface	Facilitate exploring covariates for introductory users
Merge with EOFR package by adding optional new input that is correlated with EOF axis	Facilitate exploration of spatial drivers for population dynamics
Add option to specify correlations based on traits for each category c	Likely feasible via formula interface using spatially-varying responses, but requires some investigation
Implement structural equation modelling features via covariance matrix specification
Add feature to pass through covariate names to outputs for parameter estimates and plotting	VAST now passes names for covariates to covariate-arrays in fit$data_list, e.g., fit$data_list$X1_gctp. However, it is conceptually difficult to automate the conversion of this to the parameter of coefficients, due to order and mapping used as map input. The developers welcome input on default behavior to resolve the conceptual issue.
Add options to use p-splines	Improved statistical efficiency relative to basis-splines that are currently available
Explore multi-scale correlation functions, via additive function of two Omega GMRFs operating with different kappas	Assimilate within and among-transect locational data better
Add visual diagnostics interface styled on r4ss	Allow captions to explain what is expected for each diagnostic plot
Allow distribution across categories for density/catchability covariates	In-line with alternative Joint SDM software, and useful for data-poor species
Estimate loadings-matrix among categories to be a linear function of covariates (where existing static values is a special-case of intercept-only models), perhaps starting with (L_epsilon1/L_epsilon2)[https://doi.org/10.1111/2041-210X.12723]

Completed development priorities

Please consult the NEWS document for a list of previous releases, listing features satisfying previous development priorities.