None
- Allow posterior draws of
e
to be used in predictions (with the flaguse_e = TRUE
, the new default inposterior_predict
) or generatee
based on posterior draws ofsigma_e
(obtained by settinguse_e = FALSE
)
- added a second option for obtaining predictions on new data using a previously fitted model: the options are now
predict_paramedic
andposterior_predict
. The former uses posterior draws and a Stan algorithm to generate predictions; the latter uses posterior draws and internal R functions to generate predictions. Both are expected to perform similarly.
- Updated return type of
run_paramedic
to classparamedic
; added associatedprint
,summary
functions
- added ability to obtain predictions on new data using a previously fitted model using
predict_paramedic
None
- added ability to run analysis adjusted for batch effects by setting argument
k
to be nonzero inrun_paramedic
(the default is zero for data collected in a single experiment) - added new stan code for batch analysis
- reorganized stan code into directories, since much of the code was shared between models; the new organization uses
#include
statements, and the shared code is stored indata/
(sets up data objects),parameters/
(sets up model parameters),tparameters/
(sets up transformed parameters), andmodel/
(sets up hierarchical model for shared parameters)
- removed function
run_paramedic_centered
; the desired behavior can now be obtained by usingrun_paramedic
with the argument"centered = TRUE"
- removed function
no_efficiency
; the desired behavior can now be obtained by usingrun_paramedic
with arguments"alpha_sigma = 0"
and"kappa_sigma = 0"
- removed several stan models by collapsing functionality into
variable_efficiency.stan
andvariable_efficiency_centered.stan
- updated all tests to reflect major changes
- added vignette discussing hierarchical model specification
- added a
pkgdown
site
None
- new internal functions to reduce replicating code within key functions
- new
stan
model for specifying a negative binomial distribution on V
None
- update internal
Stan
models to use log densities and unconstrained parameters, then back-transform after sampling
- allow an option to adjust for covariates
None
None
- allow
run_paramedic
withq
=q_obs
This is the initial package release!