Identified an issue when showlegend=TRUE and there is pruning.
htmlwidgets::sizingPolicy()esoph data frame would not work properly because of the value of the agegp variable, 75+. The example now recodes 75+ to 75plus.New parameter, prunebigger, prunes nodes bigger than specified size. This is particularly useful with pattern trees when there are a few unusual cases.
Vectorized the sameline parameter, to allow different behavior for different variables.
New parameter, prefixcount, to format counts (e.g. N=3).
New parameter hideconstant, is more flexible and intuitive than showuniform (also more intuitively named).
Altered vtree to work properly with Quarto.
hideconstant is specified with prunesmaller and pattern=TRUE.Small changes to the vignette.
Added new parameter thousands to specify separator (such as ",") between thousands.
Removed one color palette (RdYlGn) which is not colorblind-friendly. Also added a new colorblind-friendly palette (Greys). These palettes were pre-calculated using RColorBrewer.
Added a single-line information message when the package is loaded in interactive mode.
Added automatic text color (black for lighter-colored nodes, white for lighter-colored nodes).
Added tsummary parameter for targeted summaries.
Added showuniform parameter (TRUE by default). When FALSE, variables with no variation are not included.
Added words parameter to allow specification of a tree without any counts.
Added showrootcount parameter to control whether count is shown in the root node.
Added the.matrix dataset based on the 3 Matrix movies.
Added targeted versions of the pruning parameters, namely: tprune, tkeep, tfollow, and tprunebelow.
When ImageFileOnly=TRUE, graphics files are now numbered successively, even when not knitting.
Added justtext parameter so that justification of the extra text can be different.
Added the following options to summary: %nosort%, %freqx%, %freqpctx%, %freqpctx_%.
Fixed a small bug where the %list% summary option used digits rather than cdigits.
Fixed problem when hideconstant is specified with prunesmaller and pattern=TRUE.
Variables can now be specified as a formula with no left-hand side.
Added support for generating PDF images.
Added none: and notall: for variable specifications and summaries (these are the complements of any: and all:} respectively).
Added prefixes anyx:, nonex:, allx:, and notallx: where the x indicates that missing values will be removed.
PNG files are now automatically named vtree001.png, vtree002.png, etc. (PDF files are named vtree001.pdf, etc.)
Support was added for embedded variable trees in Sweave.
New trim parameter to support trim feature in the \includegraphics LaTeX command.
Renamed z parameter to data.
Reorganized parameters in help file and added details.
New parameter showlegendsum to show summary in legend nodes.
New parameter font to specify font to use.
Extended wildcard matching in variable specifications and summary specifications to match this*that expressions and this#that expressions.
Codes for less-than-or-equal and greater-than-or-equal were removed due to rendering issues with DiagrammeRsvg.
Fixed bug when using showlegend=TRUE with pattern=TRUE or with check.is.na=TRUE.
Fix by trafficonese for tabPanels.
Variable labels are shown differently: no longer boldface, and also larger (24 point instead of 18 point). (The metrics for boldface font seemed to be not quite right, which sometimes resulted in truncated text.)
Revamped legends.
New sortfill parameter. Specifying sortfill=TRUE fills nodes with gradient colors
in sorted order according to the node count.
vtree return value now has an info attribute,
which is a nested list that gives the count for each node
When no summary codes for statistics are specified, a default summary is shown, depending on the type of variable.
New variable prefix i: for "intersections" of variables
New variable prefixes any: and all:
New summary codes %range%, %freq%, and %freqpct%.
New variant summary codes (ending in x to disregard missing values, ending in _ to put on separate lines).
New REDCap variable prefix r: and suffix @
When vars is not specified, only the root node is shown. (Previously vtree would try to use as many variables as possible.)
Added more informative error messages.
Added additional color palettes.
The old stem: and rc: variable specification and summary prefixes
for REDCap checkbox variables are now deprecated.
The lsplitwidth parameter is now deprecated; vsplitwidth (for splitting variable names) should be used instead.
splitwidth now applies to legend nodes as well as ordinary nodes.
Fixed issue with Shiny where vtree not visible immediately.
Fixed problem with summaries when three or fewer observations are present.
Fixed bug with HTML conversion of labels in legends.
Fixed problem with DOT script when there is no root node.
Fixed bug with variable names showing up white when fillcolor="white".
Fixed bug with pattern tables when prunesmaller was specified.
New parameter, just, to specify text justification.
New parameter, arrowhead, to specify the type of arrowhead
(or none).
Added summary variable specification !=
Sebastian Gatscha added Shiny functionality
and the svg-pan-zoom JavaScript library.
The following functions were added:
init_js, inlineCssSetup, renderVtree, use_svgzoom, vtreeOutput.
New function svtree uses Shiny functionality.
Modified outputs of vtree to make it work better.
Thanks to knitr::asis_output(),
calling vtree from R Markdown when generating a PNG file has been simplified.
Specifying showvarinnode=TRUE shows variable name in each node.
When no variable list is specified,
vtree reports included variables, excluded discrete variables,
other excluded variables.
Fixed a bug with missing value nodes.
Fixed problem with return value when running chunk from R Studio.
Fixed a bug that occurred when pattern=TRUE and a single variable was specified.
Fixed a bug that occurred when showvarnames=FALSE but labelvar was set.
The keep parameter will not prune missing value nodes when vp=TRUE.
There is a new summary code, %sum%, to show the sum of values.
prunesmaller parameter to remove nodes with small counts.
In variable specifications, * matches multiple variable name endings and # matches variable names ending with numeric digits.
When called while knitting an R Markdown document, PNG files are now automatically generated and embedded.
pngknit=FALSE disables PNG generation while knitting. Returns an htmlwidget instead.
The new function VennTable reformats a pattern table output by vtree for indicator (0/1) variables.
The new function build.data.frame makes it easy to create a data frame for use with vtree.
If no variable names are provided, vtree will use, in order, all of the variables in the data frame.
Multi-variable calculations like x/y can be used in the argument of the summary parameter.
check.is.na now uses pattern=TRUE by default.
Changed the meaning of the %node% summary code,
which had previously (and confusingly!) been used to specify the variable
in whose nodes summary information should be shown.
Now %node=n% specifies node n and %var=v% specifies variable v.
Additional text produced by text or ttext or summary
now starts on the same line as the number of observations and percentage.
In order to put this text on a new line, it must begin with \n.
When seq=TRUE the first-level nodes are ordered from least frequent sequence to most frequent sequence.
(The same applies for the new pattern parameter.)
Extensive revisions to the vignette.
pattern parameter for unordered sequences of variables
ptable parameter to generate pattern tables
Variable specification modifiers: stem: for REDCap checkboxes, is.na:,
variable=value, variable<value, variable>value, tri:
choicechecklist parameter: When REDCap checklists are specified using the stem: syntax,
automatically extract the names of choices and use them as variable names.
%pct% summary code to show percentage (not frequency)
%listlines% summary code to separate items with a new line (unlike %list% which uses commas).
Summary variable specification modifiers:
variable=value, variable<value, variable>value
tlabelnode parameter: targeted node labels
ttext parameter: targeted text
New function crosstabToCases to convert crosstabulated data to a data frame of cases
runsummary parameter to control which nodes show summaries depending on node contents
retain parameter to specify additional variables that need to be available to execute the functions in runsummary
varminwidth, varminheight parameters to control node sizes for specified variables
varlabelloc parameter to control vertical justification for nodes of specified variables
This is a major release that incorporates several new features.
Automatic variable coloring using gradients.
More compact display (reduced node separation and tighter margins).
The color of the root node can be specified using the rootfillcolor parameter.
All missing-value nodes are colored white by default.
This can be changes using the NAfillcolor parameter.
By default, the root node now has no title.
NOTE Display from version 0.1.4 can be obtained by specifying plain=TRUE.
This sets:
variable-specific shades of blue
greater node separation
wider margins
missing-value nodes have the same color (shade of blue) as all other nodes at that level
showlevels is now deprecated. Use showvarnames instead.
Legends, with marginal frequencies and percentages.
To put node labels on the same line as frequencies and percentages, specify sameline=TRUE.
A new parameter, squeeze takes a value between 0 and 1,
and controls how the tightly-packed the display is.
"Sequences" of variables can be displayed by specifying seq=TRUE.
Additional summary codes:
%node=n% to show summary information only in specified node.
%trunc=n% to truncate the summary to the first n characters.
Settings for multi-way intersections can be specified with Venn=TRUE.
graphattr, nodeattr, and edgeattr
parameters allow additional Graphviz attributes to be set.
Other new parameters: palette, gradient, revgradient, singlecolor,
showpct, showlpct, showcount, varnamepointsize, lsplitwidth,
margin, showempty, prunelone