in text when not centering, added quote=FALSE to type='print' output
* Xcontrast: compute design matrices from incomplete model fit objects and predictor settings, used for rmsb::blrm pcontrast argument
Changes in version 6.7-0 (2023-05-08)
* kableExtra has been moved from Imports: to Suggests: and is used conditionally (Thanks: Dirk Eddelbuettel)
* added full likelihood ratio tests in anova() for the commonly used rms models
* reListclean: added dec attribute to better handle decimal places in case of removal of NULL or NA elements being printed; changed all model print methods accordingly
* print.ols: get n from stats, require stats to be non-empty (done for fit.mult.impute with stacking)
* LRupdate: new function to update LR test-related stats after processMI is run
* processMI.fit.mult.impute: added processing of anova result from fit.mult.impute(..., lrt=TRUE)
* print.anova.rms: added prmi argument and by default prints missing information-related parameters if anova table was created by processMI.fit.mult.impute
* prmiInfo: new function to print (or html) inputation parameters on the result of processMI(..., 'anova')
* plot.anova.rms with options(grType='plotly') in effect: fixed bug in calling plotlyParm
* added oos.loglik.* to NAMESPACE
Changes in version 6.6-0 (2023-04-08)
* latex.* for model fits: fixed case for file not blank, fixed bugs in latex.cph
* Import knitr::kable and kableExtra::kable_styling and used these in latex.cph so that math notation can appear in html tables of survival probabilities
* val.surv.Rd, calibrate.Rd: changed installed.packages to requireNamespace for speed
* impactPO: automatically call new Hmisc::combine.levels to combine adjacent levels of Y until minimum Y-level frequency >= minfreq
* processMI: new function to process Hmisc::fit.mult.impute objects, especially for averaging model validations over multiple imputations
* Removed package dependencies except for Hmisc
* Exported survival::Surv function so users will not have to load the survival package just to access Surv; likewise for ggplot so it's easy to do ggplot(Predict(fit))
* npsurv: re-wrote to recognize that survfit is imported, not attached
* html.na.print.delete: changed from using plotly to using Hmisc::dotchart3 converted to base64 image so that will render correctly in Quarto reports
* predab.resample: respected prmodelsel in not printing title
* residuals.lrm, orm: Chun Li greatly improved calculation speed for Li-Shepherd residuals
* fixed bug in Hmisc::dotchartp that made plot.anova.rms not work with changes in R in handling of matrix[NULL,]
Changes in version 6.5-0 (2023-02-06)
* modelData: preserved row.names so that fitted values, residuals will have them
* ols.influence: added row names back to dfbetas (makes which.influence work)
* latexrms: changed to return a character vector if file != ''
* latex.*: changed to use knitr::asis_output if options(prType=x} is in effect where x is not 'plain', and if file is not ''. This will result in correct rendering in documents no matter which chunk options are in effect, i.e., you no longer need results='asis' in a chunk header to get the LaTeX math form of fitted models.
* print.*: likewise for printing model fits, anova, summary, validate
* latex.anova.rms: added which= for html and LaTeX
Changes in version 6.4-1 (2023-01-22)
* latex methods: changed \text to \mathrm; was getting a mathjax syntax error in Quarto
* residuals.ols: added type="studentized"
* poma: merged https://github.com/harrelfe/rms/pull/123 with thanks to Yonghao Pua
Changes in version 6.4-0 (2023-01-12)
* bootcov: added seed argument; need to change old uses to use this; affects usage of the boot package
* html. and latex.anova.rms: added fontsize argument
* plot.summary.rms: added declim argument
* ggplot.Predict, plotp.Predict, plot.summary.rms: override plotly height and width if options(plotlyauto=TRUE) is set
* html.validate: fixed call to htmlTable to not add special css as was getting an html error
* latexrms: for inline use array environment with double backslash as line break
* latex.lrm, latex.orm: changed to used standard LaTeX probability notation
* poma: fixed by Michał Krassowski - https://github.com/harrelfe/rms/pull/116
* html.naprint.delete: added

for heading for html output
* impactPO: fixed printing for case where not all models are run
* latex.cph: use standard probability notation and $$
* latexrms: changed from eqnarray* to array, changed math notation back to $ for spline and indicator notation
* all latex methods: changed to output markup as we go, $$ for display math
* residuals.Glm: new function, adds type='score' for use with robcov
* makepredictcall.rms: new function that makes rms transformation functions like rcs, lsp, gTrans give proper predicted values when non-rms fitting functions are used and standard predict methods are implemented for thos functions. Thanks: Terry Therneau
* orm: new example in help file for plotting different exceedance probabilities from a fit
* impactPO: fixed for case where bootstrap is requested, by changing data to data frame if an environment
Changes in version 6.3-0 (2022-04-12)
* gTrans, Design, set.atr, latexrms: For latex(fit) to work when there is a gTrans() predictor
* survreg.fit2: fixed bug in computing loglik transformation correction factor which ruined validate.psm R^2 and other loglik-related indexes (made them NaN)
* residuals.lrm, residuals.orm, predab.resample: fixed bug where x=TRUE in fit without y=TRUE may find another y object; thanks: Chun Li
* plot.summary.rms: for options(grType='plotly') changed height calculation to use heightDotchartb instead of heightDotchart
* residuals.ols: added type='influence.measures'
* added nnet to suggests
* added impactPO function
* added new adjusted Maddala-Cox-Snell adjusted R2 using new Hmisc function R2Measures; affects prModFit, orm, lrm, psm, cph
* adapted reListclean in rmsMisc to handle special element namesFrom and named vector elements
Changes in version 6.2-0 (2021-03-17)
* plotp.Predict: fixed bug when conf.int=FALSE was given to Predict()
* gTrans: new general transformation function; example: linear spline with discontinuities
* validate.cph: surrounded survival::survConcordance with suppressWarnings
* fixed how matrx predictors were being handled
* contrast.rms: added better ggplot2 examples for odds ratios
Changes in version 6.1-1 (2021-02-06)
* Design, Getlim: allowed datadist option to actually contain the datadist (used by Hmisc::estSeqMarkovOrd)
Changes in version 6.1-0 (2020-11-28)
* stackMI: removed and moved to rmsb package
* Predict: changed name of first argument from x to object
* Changed email address
* modelData: fixed dropping of unused factor levels
* contrast,summary,Predict: changed interface with rmsb::blrm to use more of rmsb::predict.blrm and made summary.rms and contrast.rms handle contrained partial PO models and added ycut argument
* poma: new function by Yong Hao Pua