apeterson91/rstapdp

Functional Spatial Temporal Aggregated Dirichlet Process Predictors

rstapdp's Issues

Keywords Documentation

• Add documentation on sap(),tap(), or stap() keywords used in formulas.

• Clarify that the current implementation only allow for use of one such term and clustering difficulty often grows in proportion to the dimension of the smoothing covariates.

• -- Consequently a tensor product term (e.g. stap()) may have more difficulty clustering than a single smooth term (e.g. sap())

Add functionality for multiple/parallel chains

See rstan's implementation using mclapply for unix systems.
May have to restrict use to unix systems as first step since the windows set up looks more complicated

Testing issues

testthat errors when using code that captures output: e.g.
capture_output( lmer1 <- fdp_staplm(BMI ~ sex + sap(FFR),benvo=FFR_benvo))
Need to fix this so that tests incorporating model fits can be run.

Terms function results in inappropriate S matrix in case where there are dummy variables

Spline clustered functions

Current datasets use clusters of parametric functions, typically weibull(). Create a new dataset with clusters spline function, in a longitudinal context.