- fix to shiny.hdlm and shiny.hdlmm to remove bug where no data is selected
- fix to combine.models.R to correctly combine tree posterior samples
- fix to summary.tdlmm in rowMeans to stop autodrop of dimension
- a split point for Subgroup tab in shiny interface set to the mode of proportion
- initial CRAN release
- spelling checks using
spell_check_package()
function fromspelling
package - S3 method
predict.dlmtree
is separately defined aspredict.hdlm
andpredict.hdlmm
- combined model fitting functions
tdlnm
andtdlmm
intodlmtree
- Beta prior for monotone CW selection
- Bayes factor selection method
- zero inflated tree method
- monotone DLNM
- precalculation in tdlnm for large speed improvement
- fix to single covariate model
- add functions for posterior analysis with dlmtree class (
pip
andsplitpoints
)
- fix to allow intercept only models
- updated tdlnm algorithm for increased performance (memory restriction for large datasets)
- add HDLM functionality (
dlmtree
)
- fix
summary.tdlmm
if nMix=1 or nExp=1
- combined identity and logistic link into single c++ file
- several code efficiencies
- updated print output of
tdlmm.summary
- removed prior on dirichlet hyperparameter
- exposure selection based on Beta-Binomial distribution
keep.mcmc
in summary
- separated
tdlmm
fromtdlnm
function - included exposure selection into summary
- prior on dirichlet hyperparameter ~Gamma(1, nTrees/4)