Contains many functions useful for data
analysis, high-level graphics, utility operations, functions for
computing sample size and power, simulation, importing and annotating datasets,
imputing missing values, advanced table making, variable clustering,
character string manipulation, conversion of R objects to LaTeX and html code,
and recoding variables.
Harrell Miscellaneous
Current Goals
Continue to refine the summaryX class of functions that replace tables with graphics
Consider using the haven package for importing SAS, Stata, and SPSS files; haven stores labels as the label attribute of each variable as does Hmisc; it converts date and time variables automatically and allows one to specify a format catalog along with the primary dataset
See if the readstata13 package has advantages over the foreign package for Stata file import
Consider creating xl.get using the readxl package to read .xls and .xlsx Excel files
In impute.transcan, sense if a variable in data is not a factor whereas it was treated as a factor during aregImpute; it should be converted to factor before the line v[sub] <- ... levels(as.integer...)) is run
News
Changes in version 4.2-0 (2019-01-25)
putHcap: new function
inst/tests/example.Rmd: changed to html document, replaced putHfig calls with new putHcap
knitrSet: new lang='blogdown' option
fit.mult.impute: suppressed warning for orm if there were no missing Ys (thanks: Lucy D'Agostino McGowan)
knitrSet: made more functional for blogdown
knitrSet: removed duplicate setting of out.width
fit.mult.impute: added step-by-step instructions for imputing missing baseline variables when analyzing serial data
histboxp: changed to display quantiles using line segments, and centered segments vertically. added connect argument with default set to connect quantiles with a line
wtd.quantile: removed observations with missing weights
print.summaryM, latex.summaryM: if no statistical results, do nothing and return
naclus: checking missings in character variables to also use is.na
dotchartpl: backsolve from Wilson confidence interval when prop=0 or 1; added nobigtracename argument; fixed trace names
putHfig: added file=FALSE to return character vector instead of writing
histboxp: added showlegend so that can be called in a loop
histboxpM: new function to call histboxp for multiple variables in a data frame
htmlSN: re-wrote for speed, and added argument pretty=
plot.summaryS: added plotly method when options(grType='plotly')
plotp: moved generic from rms to Hmisc
plotlyM: new function to make it easy to use plotly to produce multiple graphics of certain types, with specifications coming from a data frame
plotp.summaryS: new plotly method for summaryS
mbarclpl, medvpl: new auxiliary functions for plotp.summaryS
dhistboxp: new function
markupSpecs$httml: added new function mdchunk to intersperse markdown with R object output
markupSpecs$html$varlabel: if html is TRUE run label and units through htmlTranslate
dotchartpl: removed stratified by groupname from hover text
plot.summaryS: removed transhtml argument
markupSpecs$html cap-type functions: called htmlTranslate(...,greek=TRUE) on long and short captions
formatTestStats: ran P-values with < through latexTranslate to make latex.summary.formula.reverse and latex.summaryM work correctly
hidingTOC: new function by Thomas Dupont for various features for floating table of contents with RMarkdown html reports
histbackback: use pretty() to label breaks
cut2: added formatfun argument, by github.com/harrelfe/Hmisc/commit/c8816d589b2e95eed15b96a886558a92bb143abd
latex.default: checks options(omitlatexcom) and if TRUE suppresses emitting LaTeX comments; keeps Rmarkdown/pandoc from trying to escape LaTeX comment characters, putting the comments in the final output
dotchartpl: added refgroup, sortdiff, conf.int arguments to allow inclusion of difference in two proportions across levels of the group variable
plot.summaryP: for plotly, added refgroup to pass to dotchartpl so that differences in proportions can be handled; also passed ... arguments
dotchartpl: added minkeep argument (used by hreport eReport)
markupSpecs$html: added function uncover for hiding chunks; fixed typo in definiton of larger function
knitrSet: when lang='markdown' sets up hook option uncover
knitrSet: added aliases ho for out.height, wo for out.width, added knitrSet arguments ho, wo
wtd.var: replaced with fixed verion by Benjamin Tyner [email protected]
inst/tests/example.Rmd: added example with simple caption and font size change for caption
pomodm: new function to work with popower/posamsize to assist translating odds ratios to means and medians for proportional odds model
histboxp: added wmax, mult arguments to tweak margin set-aside
describe: fixed bug in html.describe when there is only one variable
simRegOrd: new function to simulate covariate-adjusted ordinal model two-sample test power
html: use htmlSpecialType for functions calling htmlSpecial, htmlGreek
htmlSpecialType: new function to check options('htmlSpecialType') and use as a default '&' for special symbols
htmlSpecial: added new characters, cleaned up code
impute.transcan: if original variable was logical, convert 0/1 imputations to FALSE/TRUE; design matrix column names were problematic with fit.mult.impute in this case
impute.transcan: converted character variables on input dataset to factor so can replace NAs with imputed codes
htmlTranslate: fixed bug when unicode was in effect: characters were shifted;fixed &# code for pound sign
errbar: properly passed arguments main and sub to plot
html markupSpecs: added sectionNumberDepth function
wtd.quantile: remove zero weight observations early
latexCheckOptions: new function from Richard Heiberger
htmlSN: used new HTML specials mechanism for times symbol
summaryM: added escape.html=FALSE to htmlTable calls
combine, combine<-: removed (were deprecated long time ago)
removed inc-dec.s which defined unused inc<- and dec<- for increment and decrement
NAMESPACE: removed exportPattern("^([^.]|\..+\.)") and explicitly exported everything, stopping export of methods
round.POSIXt,trunc.POSIXt: changed to roundPOSIXt, truncPOSIXt to avoid conflicts with R base
asNumericMatrix, matrix2dataFrame: changed to respect original storage.mode. Thanks: Andre Mikulec
Changes in version 4.0-3 (2017-04-30)
bpplot: added plotopts argument
describe, histboxp: improved logic to still slighty round numeric variables with fewer than 100 distinct values, in case some values are really close, when computing frequency distribution (for spike histograms in plot.describe)
all.is.numeric: fixed to handle case where all values are " "; makes describe work
latex.describe.single: changed to always issue vbox, to improve vertical spacing
histboxp: removed zero line so wouldn't collide with data
ggfreqScatter: added smoother to example
latex.summaryM: more transparent naming of quantiles preserving full precision
cut2,describe,dotchart3,ecdf,latex,list.tree,bpplt,panel.plsmo,pstamp,summary.formula,summaryM,transace,t.test.cluster,wtd.Ecdf,Dotplot,zoom: fixed on.exit to properly restore options() to original state
[.pBLock: fixed old behavior of only rataining the pBlock class if the result is a matrix (thanks: Michael Lawrence - lawremi)
addMarginal: added nested argument
putHfig: now throw error if lcap is given with expcoll because of interaction with divs in markdown sub-subheading
plotmathTranslate: improved to put any non-legal expression inside paste()
plot.summaryS: separated xlim and ylim to fix bug where xlim was ignored
plsmo: stop on error if no. observations not large enough for m with method='intervals'
Fortran calls: registered all Fortran functions, removed package name and quotes in first argument in .Fortran(), added F_ to entry point name in those calls, changed useDynLib call in NAMESPACE, added \alias{} for all F_ entries
Changes in version 4.0-2 (2016-12-30)
pngNeedle: fixed bug in html.describe under Windows by not specifying /tmp as the temporary file directory - https://github.com/harrelfe/Hmisc/issues/63; used tempdir() instead of tempfile()
html.describe.single: reuse the same temporary file for intermediate png output for spike histograms
htmlGreek, htmlSpecial: new internal functions
htmlTranslate: added unicode argument, default FALSE, use htmlGreek
tobase64image: new internal function. Thanks: Dirk Eddelbuettel
Changed IMPORTS: base64 to base64enc
spss.get, stata.get: made use of foreign package files' ability to use URLs as the file argument
sasxport.get, getZip: allowed for https (thanks: Kurt Hornik)
getRs: removed unused where argument
latex: revert changes that removed fix to column duplication in cellTexCmds
latex: added continued argument
plot.describe: fixed bug in not using formula notation for plotly
histSpikeg: added plotly support, stop curtaining x range when xlim omitted, use approxExtrap to show data density outside of prediction range for x
histSpikep: deleted function
dotchartpl: added lower, upper
describe.vector: changed structure(NULL, ...) to structure(list(), ...)
Changes in version 4.0-1 (2016-12-04)
gbayes: new test file, fixed bug in posterior mean when mix=0,1 in gbayesMixPost
ggfreqScatter: added prfreq and cuts arguments
latex.summary.formula.reverse, latex.summaryM, html.summaryM: fixed bug with backslash added before % for html, improved typesetting language logic
latex.describe: fixed bug that prevented spike histogram from appearing
ggplot.summaryP: moved text out of aes in geom_point, left in main aes
plot.describe, plot.summaryP: put text aes in ggplot(), increased ggplotly height to allow popups for top continuous variables in plot.describe
dotchartp, dotchartpl, plot.describe: moved height and width from plotly::layout to plot_ly because of changes in plotly 4.5.6
bppltp: added data= to plot::add_* because plotly was evaluating using the wrong copy of x,y, etc.
plot.summary.formula.reverse, plot.summaryM: removed subtitles argument since not used
upData: fixed bug when converting existing variable to factor; was not recognizing new storage model and tried to run floor()
latex.summary.formula.reverse: fixed bug where outer quantiles and frequencies were not being rendered
latex.default: put \par in front of insert.bottom if table.env is FALSE, so that text is not put to the side of the table
histboxp: new function for plotly stratified spike histograms with box plot statistics also depicted
format.df, latex.default: added arguments already.math.row.names, already.math.col.names that if TRUE prevent any math mode changes
latex.default: fixed problem with duplication of columns in the cellT...exCmds array (thanks: Thomas Dupont)
ggfreqScatter: better formatting of axis tick mark labels
latexVerbatim: implemented file=''
Changes in version 4.0-0 (2016-10-31)
summaryP: fixed exclude1 logic - was not excluding enough levels (inconsistent use of variable names vs. labels)
latexTranslate: any NAs in first argument changed to "" before conversion
minor.tick: added arguments to pass through (thanks: vhorvath)
tests/latexpng.r: test conversion of LaTeX table to png
latexTabular: made it properly call latexTranslate separately for each column, convert argument to matrix or data frame if a vector
tests/latexTabular.r: new test
latexDotchart: added call to latexTranslate to escape special characters for LaTeX such as underscore and pound sign
ggfreqScatter: new function
grType: new non-exported service function to sense if plotly is in effect
plot.describe: new function to make plotly graph for describe objects
describe: changed output: 3 decimal places for Info, new format for frequency table, separated logic for 10 extreme values from logic for frequency table, significant changes to print.describe
dotchartp: new version of dotchart3 for plotly charts
summaryD: modified to use dotchartp if options(grType='plotly') is in effect
nFm: added argument html
label.default, label.Surv, labelPlotmath: added html argument to implement HTML markup in place of plotmath (for plotly)
now imports htmlTable, viridis
html.contents.data.frame: return invisibly if file=''
htmlVerbatim: new function
html.contents.data.frame: improved html
html.data.frame: changed column headers from h4 to
ggfreqScatter: added html argument
knitrSet: was ignoring default figure h and w
html.summaryM: new function using new version of latex.summaryM
markupSpecs: new list object defining markup elements for latex and html, used by summaryM
show.html, print.html: removed; conflicted with htmltools/rmarkdown
label: required exact match on attribute name "label" to not retrieve the "labels" attribute (thanks: Ivan Puzek)
htmlSN: new function to convert floating point to scientific format usint html
upData, cleanup.import: fix NA becoming new factor levels (thanks: Beau Bruce)
htmlTranslate: new function
plotlySave: new function
histSpikep: new function
upData: new argument html
plot.describe: added ggplot2 graphics, improved continuous display using ggplotly
labelPlotmath: cleaned up by using markupSpecs
capitalize: fixed bug - did not cap. first letter if other letters after it were upper case (thanks: dchiu911)
html.summaryM: added brmsd argument to put mean, SD on separate line
Save: changed default back to use gzip for speed
knitrSet: used new knitr way to set aliases for option names
latexTabular: made translate argument apply to body of table also; implemented multi-line header
html.data.frame: added several arguments
describe: added html method
html markupSpecs object: added bibliographic database utility functions
bppltp: new service function for extended box plots for plotly
plot.summaryM: new plotly method
prType: new service function used to detect user settings for html, latex for print methods
html.contents.data.frame: removed file and append arguments and output an html character vector instead
prList: new function
GiniMd: moved to Hmisc from rms
describe: added GMD (Gini's mean difference) and relabeled unique to distinct
latex.summaryM: added Needspace{2.7in} before 2nd and later strata
ggplot.summaryP: added point labels for use with plotly as hover text
latex.summaryP: fixed bad linebreak at strata boundaries
dotchartpl: new plotly dot chart function especially for summaryP
plot.summaryP: new plotly method using dotchartpl when options(grType='plotly')
putHfig: new function to render graphics with captions in HTML reports
pngNeedle: new function, like latexNeedle but useful with HTML reports
html.data.frame: added width and caption arguments
plotlyParm: list object with plotly helper functions
summaryD, dotchartp: added symbol and col arguments
upData: improved efficiency, added labelled class to variables without it but having labels (e.g., data imported using haven package)
gbayesMixPost: mixed error in posterior odds of the mixing parameter that resulted in incorrect posterior probabilities when the variance of the statistic was appreciable; added option to compute posterior mean
Changes in version 3.17-4 (2016-05-02)
latex.summaryP: correctly ordered rows after running reshape because groups was present
rcspline.plot: took confidence limits into account when computing default ylim
bootkm: new test in tests
[.ynbind: Made it return an unclassed vector if the user was subsetting to a single column. Upcoming R release fails in table() otherwise. Thanks: Michael Lawrence [email protected]
Merge: added function (moving from greport package)
data.table: imports
Changes in version 3.17-3 (2016-04-03)
latex.default: added blank lines after textbf with insert.top unless center='center' (was unless 'none')
plot.summaryS: fixed bug were limits were not aligned with columns (thanks: Liping Du)
mgp.axis: respected par('tcl') but only as extreme as -0.4
mgp.axis: added cex.axis, cex.lab arguments
aregImpute: added reformM function for permuting order of variables. Thanks: Yong Hao Pua
latex: fixed rownamesTexCmd error
Changes in version 3.17-2 (2016-02-13)
consolidate: A simpler version of assignment form of consolidate, avoids eval/match.call and works with the byte-compiler (thanks: Tomas Kalibera)
relevel.labelled: new function (thanks: Max Gordon)
fit.mult.impute: remove contrasts attribute for imputations done by mice
formatCons: fixed bug when stats contains one row (thanks: Cole Beck)
Hmisc.h: removed reference to S.h
fit.mult.impute.s: moved source code to separate file, away from transcan.s
fit.mult.impute: added warning if fitter is lm, which does not use imputation variances for sigma or standard errors
spss.get: added reencode argument (Thanks: Andreas Osowski)
Changes in version 3.17-1 (2015-12-17)
sasxport.get: added lowernames argument with default equal to the previous behavior (convert varixcable names to lower case)
biVar, rcorrp.cens, spower, transace: changed 1 - pnorm(z) to pnorm(-z)
latex.summaryM: added arguments prob, prN, and legend.bottom
print.summaryM: added arguments prob and prN
Save: changed to use compress='xz' by default, and added compress argument
drawPlot: fixed bug in defining Points and Curve functions; added col argument to Points and Curve; removed helper functions for Points and Curve from Key.s; returned object invisibly; added type='loess' to Curve()
upFirst: new function
mdb.get: added mdbexportArgs argument which among other things allows -b strip to be omitted
html: for hevea added definition of tabularnewline as \, changed method to default to hevea
summaryM: fixed calculation of N for mChoice variables
stat-plsmo.r: re-written by Hadley Wickham (thanks!) to work with new ggplot2
ggplot.summaryP, ggplot.transcan: added mapping and environment arguments to satisfy rules about generics
Changes in version 3.17-0 (2015-09-20)
format.df (used by latex.default): added space after textless, textgreater
label: changed default for units to value of plot
getRs: replaced where argument with guser, grepo, gdir, dir to allow easy fetching of updated functions from Hmisc etc.
Separated sas.get source code from other .get functions and from upData/cleanup.import by putting into 3 separate files. Moved stata.get into misc.get.s
upData: for Stat/Transfer exported R workspaces, change variables into factors to incorporate value labels when present; added subset argument and reporting of number of observations pre and post subsetting
latex.default: added comma after botcap directive for ctable. Thanks: Paul Trowbridge
Hmisc-internal.Rd: removed alias{[.terms}
latex.default: for longtable when no caption is given, subtract one from table counter
latex.summaryM: quit ignoring insert.bottom if it is a character string (thanks: JoAnn Alvarez)
knitrSet: new function to setup knitr with nice defaults for books etc.
rcorr: fixed sensing of NAs and diagonal elements of n matrix; thanks: Keith Jewell, Campden BRI Group; similar for hoeffd
Changes in version 3.15-0 (2015-02-15)
dvi.latex: For Windows switched from system() to shell() so that cd command works. Thanks: Rich Heiberger
histSpike: added test
histSpikeg: new function for ggplot2
added ggplot2 in DESCRIPTION, NAMESPACE
largest.empty: clarified help file for use with ggplot2, changed how sensed xlim and ylim accordingly
stat_plsmo: new geom for ggplot2
transcan: added ggplot method
transace: removed y-axis labels
residuals.lrm: remove y-axis labels
histSpikeg: added lowess and span arguments
format.df: Andreas Kiermeier: The "..." argument of format.df is no longer ignored, but included in the definition of formt() and passed to format(). The help file has also been updated and the example modified to show an example of big.mark.
getHdata: removed S-Plus specific stuff, updated url to DataSets
summaryP: added ggplot method
NAMESPACE: many additions to register all S3 methods
format.sep: renamed to formatSep
Changes in version 3.14-6 (2014-11-16)
wtd.rank: fixed bug in no weights case (thanks: DPlat)
latex.summaryM, latex.summary.formula.reverse: added npct='slash' to present numerators and denominators horizontally instead of vertically
plsmo: put derivation of ylab earlier so will use original attributes
latex.default: changed to use colheads=FALSE to suppress column headings. Thanks: Michael Rose
mdb.get: added system option -b strip to mdb-export to skip binary output
describe help file: added note that correct terminology for "unique" should have been "distinct"
Changes in version 3.14-5 (2014-09-11)
latex.summaryM: fixed bug in caption with test=TRUE. Thanks: Yonghao Pua
csv.get: uses data.table package's fread in place of read.csv if data.table is in effect
combined.levels: sensed all NA vector, now return non-factor numeric instead
dataframeReduce: handle all-NA factor variable
subplot: replaced with latest version from TeachingDemos package by Greg Snow
latexTabular: fixed error in example in help file; result is not a file
latex: another test added in tests/latex.s
summaryP: removed observations with a right-hand-side variable missing
latex.summaryP: fixed bug with wrong column labels due to reshape reordering columns coming from factor levels alphabetically instead of by original levels
format.df: added % & <= >= to list of characters handled, the last two by going into math mode
latex.summaryP: use blank if denominator 0, instead of NaN
summary.formula: fixed problem with deparse formula. Thanks: Andreas Kiermeier
describe: added relative information measure for numeric variables - a measure of how continuous the variable is
wtd.table: detect duplications using duplicated() instead of diff(x) to handle Inf. Thanks: Benjamin Tyner
DESCRIPTION, NAMESPACE: multiple function changes to work in R-devel
Changes in version 3.14-4 (2014-04-13)
rcspline.eval: stop new logic for ignoring outer values when there are many ties when there are also many ties on interior values. Added new logic to use interior unique values of x when the number of unique x is small.
latexBuild: generalized with insert argument
latex.default: modified to use mods of latexBuild, fixed bug with caption.loc='bottom' (thanks: YacineH)
latex.default: fixed bug where comma did not appear after caption={} for ctable (thanks: Johannes Hofrichter)
tests: fit.mult.impute.bootstrap.r: added new example (thanks: Jane Cook)
fit.mult.impute: added call for fit.mult.impute in returned object, replacing call from fitter; makes update work for fit.mult.impute
summary.formula: fixed recognition of multiple left-hand-side variables to trigger call to summaryM (thanks: Kevin Thorpe)
summaryM: changed group label to '' instead of 0 for formulas like age + sex ~ 1
Ecdf: added what argument to all functions
nobsY: return constructed id vector
addMarginal: instead of .marginal. being logical, make it contain names of variables being marginalized over
mChoice.c: fixed some inconsistencies
Changes in version 3.14-3 (2014-03-02)
format.df: clarified in help file that col.just can contain specifications other than l,r,c, e.g., "p{3in}" to get paragraph formatting in a column. Thanks: Ben Bolker
latex.default: added example for the above in tests
label.Surv: got units from inputAttributes in addition, and added type argument
Changes in version 3.14-2 (2014-02-26)
latex.default: improved logic using new function in Misc: latexBuild
latex.default: fixed bug with ctable=TRUE with no caption by removing default label
latex.default: improved formatting for insert.top
latex.default: added tests, fixed insert.bottom
latex.summaryM: return stat summary key as legend attribute, use this according to insert.bottom argument
latex.summary.formula.response: fixed bug related to computation of cdec. Thanks: Kevin Thorpe
latex.default: added new argument star: ctables uses this to spread over two columns when the LaTeX document is in \twocolumn mode. Thanks: David Whiting
Changes in version 3.14-1 (2014-02-25)
Added latexNeedle function
Change latexTherm, latexNeedle to use user LaTeX macro \tooltipn to do the pop-up
latex.default: changed line breaks around \end{tabular}
latex.summaryM: put insert.bottom text in minipage so \tooltip will not devote wide space to it
sas.get: added defaultencoding argument and logic (Thanks: Reinhold Koch)
plot.summaryP: omit tick marks for proportion > 1.0
format.df (used by latex): fixed na.blank logic for character var
latex: removed newlines when ending environments, added hyperref argument
latex.default: changed 'rotate' to 'sideways' for ctable mode. Thanks: Simon Zehnder [email protected]
gView: removed
ldBands: removed
summaryP: new function - graphical replacement for tables of proportions
ynbind: new function for combining related yes/no variables into a matrix with a label
added file argument to prn
summaryP: added autoarrange
added addMarginal and nobsY functions
pBlock: new function for blocking variables for summaryP
summaryP: changed text positioning to grid absolutes, added text.at argument
scat1d, histSpike: if grid used and y has grid units, fixed logic for frac
plsmo, panel.plsmo: added scat1d.opts argument
label.Surv, units.Surv: added, removed ::: in survival calls
summarize: added keepcolnames argument
Suppressed startup message unless options(Hverbose=TRUE) is set
summaryS: new function - multi-panel lattice xy and dot plots
summaryD: added ylab argument
dotchart3: quit letting left margin be less than pre-existing one
multLines: new function
Improved nobsY to respect subject IDs when counting number of subjects, and to return an attribute 'formula' without id variable; changed bpplotM, summaryP, summaryS to use this
Removed nobsY calculations from bpplotM, summaryP, summaryS, enhanced nobsY to allow stratification by treatment
panel.bpplot: added violin and violin.opts arguments
summaryS: added medvPanel support during-plot vertical violin plots
plot.summaryP: padded x-axis limits
latexTabular: added translate and hline arguments; moved to its own file and help page
latexTherm: added tooltip using LaTeX ocgtools package
summaryP: stopped reversing order of panels
summaryM: added table.env argument, changed how model.frame built
latex.summaryM: changed to print proportions by default, added round='auto'
character.table: added xpd=NA; thanks: Dale
summaryP: added latex method
latex.default: added insert.top argument
summaryM: added stratification (multiple tables)
Changes in version 3.13-0 (2013-11-18)
Changed n location (nloc argument) in bpplotM
Improved dotchart3 to better compute string widths when there is a mixture of expressions and regular strings for auxdata/auxtitle
Changed rlegend to not take logs if log axes are in effect. Fixes Ecdf(..., log='x', label.curves=list(keys=1:3)). Thanks: Bayazid Sarker [email protected]
Extended non-panel (regular) version of plsmo to handle matrix y
Likewise for summaryRc
Added xlim to bpplotM
Added latexTherm function to create LaTeX picture environments to add a series of thermometers to LaTeX text
Fixed deff to handle the case where R^2 = 1. Thanks: Matthieu Stigler [email protected]
Added new test file for wtd.mean, wtd.quantile
New test aregImpute3.r for glm Poisson regression
Improved describe.vector to format single unique values
Took away warning about var, s.e., t, p in fit.mult.impute
Switched from subversion to github repository
Changed maintainer from Charles Dupont to Frank Harrell
Changed wtd.loess.noiter to use loess instead of stats:::simpleLoess