NEWS
Update History for Package 'lessR'
-------------------------------------------------------------------------
------------------------------------------------------------------------
------------------------------------------------------------------------
>>>>>> Changes for lessR version 3.4 (2015-12-27) <<<<<<
-----------
Updates 3.4
-----------
error trapping
more development of lessR error trapping to replace the more cryptic
R error messages with more understandable messages that also provide
guidance as how to correct the problem
of existing lessR explanations and the following additions:
1. specifying variables to analyze that do not exist in the data table
2. specifying variables to analyze without having a data table
3. naming the intended default data table Mydata instead of mydata
4. calling a data frame in place of a variable in ttest
5. improperly enclosing a variable name in quotes in a function call
6. failing to specify a variable to analyze in ttest and ScatterPlot
7. trying a histogram for a categorical variable
8. trying a scatter plot with the second variable non-numeric
BarChart
colors changed for two variable plots, now based on hues generated by
rainbow_hcl(24,c=38,l=75) from the colorspace package, such that
when desaturated all colors have the same shade of gray
proportions option now available for 2 variable plots
phi coefficient or Cramer's V displayed with two variable analyses
PieChart
frequency distribution added so that text output is same as BarChart
colors same as BarChart for two variables
SummaryStats
chi-square test provided
phi coefficient or Cramer's V displayed with two variable analyses
Help
spacing improved with shorter lines of output
Logistic
collinearity analysis restored
Regression
subsets option can be an integer to specify maximum number of lines
displayed, where each line represents a specific subset model
for subsets of more than 40 lines, the variable names are written
each 30 lines
scatter plot matrix adjusts the size of the correlation coefficients
depending on the number of predictors
better labeling of subsets output to indicate that only the best
10 models of each number of predictors are considered, when relevant
Read, details
More concise output
Hadley Wickham's read_excel function restored for reading Excel files,
and Read re-interprets the variable types from read_excel so that
they are equivalent to those from reading other file formats with Read
ttest
for paired analysis, difference score now computed from subtracting
first variable from second variable
ScatterPlot
ellipse option restored, with axes automatically reset to provide for
values of the ellipse that exceed the range of values of the data,
and fill.ellipse color can be specified, usually with partial
transparency such as rgb(.8,.8,.8,.2)
correlation analysis restored
if xy.ticks is FALSE, then axis labels moved closer to the plot
ellipse possible with bubble/sunflower plot
VariableLabels
new function that essentially replaces label function, with new
features of reading a file of variable names and labels separately
from the Read function, and also from the console
-------------
Bug Fixes 3.4
-------------
BarChart
legend printed in light text if background is dark
Histogram, SummaryStats
outlier analysis for small outliers improved
Merge
variable units now properly processed
ScatterPlot
empty graphics window no longer generated
SummaryStats
when number of unique values <= n.cat, properly treat variable
as categorical
ttest
missing data allowed for paired version
>>>>>> Changes for lessR version 3.3.6 (2015-11-05) <<<<<<
-------------
Updates 3.3.6
-------------
ANOVA
output now constructed in segments for better knitr compatibility,
> a <- ANOVA( ... )
> a # view all the output
> names(a) # view the names of the segments
> a$out_anova # for example, view the summary table
knitr.file option for automatic construction of markup file
from the various output segments
improved formating of summary table
graphics=FALSE option added
----------------------------
lessR Installation Bug Fixes
----------------------------
Read
because of potential package dependency problems loading packages
with the readxl package function for reading Excel files, went
back to the gdata package for reading Excel files, which requires
Perl, which requires a download for Windows computers, and which,
unfortunately only reads the formatted data not the actual data,
so first format the Excel data according to the General format
before reading
ScatterPlot
because package dependency problems loading packages with the car
package, the ellipse option from that package is deactivated
Regression
because package dependency problems loading packages with the car
package, the scatter.3D option from that package is deactivated
---------------
Bug Fixes 3.3.6
---------------
Regression
background, which listed variables in the model, sample size, etc.,
displays the intended information
>>>>>> Changes for lessR version 3.3.4 (2015-08-22) <<<<<<
-------------
Updates 3.3.4
-------------
Read, details
better display format of variable labels and units
Regression
knitr.file option has added display options code for displaying the
code that generated the results and document for documenting the code
knitr.file option extended to work with mydata <- rd(), that is browse
for the file to read before doing the regression
improved use of variable units in R Markdown from knitr.file option
Set
can set the values for display options globally in the generated
knitr.file generation, which includes results, explanation,
interpretation and document
---------------
Bug Fixes 3.3.4
---------------
ANOVA
missing terms from the sums of squares table included
Histogram, Density and BoxPlot
variable names that are also function names are properly processed
Merge, Subset
variable units preserved
Read
specified format of data file remains regardless of file type
>>>>>> Changes for lessR version 3.3.3 (2015-07-23) <<<<<<
-------------
Updates 3.3.3
-------------
Read
Hadley Wickam's read_excel function used for reading excel files,
which does not require Perl, with the character variables from
read_excel set as factors as with reading other data formats,
except for the following addition for all formats ...
Non-numeric character strings with unique values read as class
character instead of class factor
Regression
extensive further development of the generated markup file from
the knitr.file option
up to 6 predictor variables allowed for specifying new data,
instead of just 5
Logit
up to 6 predictor variables allowed for specifying new data,
instead of just 5
corCFA
knitr.file option
option for lavaan style model specification, the same code runs
both corCFA and lavaan
min.cor and min.res options added for minimum respective value to be
printed, to improve readability of correlation matrices
output correlations omit the decimal point for more compact output
correlations predicted from the model available in assigned
output of function
factor labels displayed on output correlation matrices
Label, Details
print formatting of labels improved
Read
SAS files are read
Write
specify parameters in a more standard order: ref, data, format
---------------
Bug Fixes 3.3.3
---------------
corRead
abbreviated form rd.cor properly recognized
>>>>>> Changes for lessR version 3.3.1 (2015-04-27) <<<<<<
-------------
Updates 3.3.1
-------------
BoxPlot, Density, Histogram
output generated with named pieces such as for knitr,
plus knitr.file option
Regression
knitr.file much further developed including reproducing the full function
call to the Regression function where the knitr.file is created, and
now includes flags for output control: explain, interpret, results
knitr information is now only written to a file, not to the
output object
displayed prediction intervals always contain the smallest interval
and the largest interval
PRESS R-squared included in the default output
minimum number of decimal digits on output changed from 3 to 2,
e.g., integer input leads to 2 decimal digits by default (override
with digits.d)
spacing of tabled output condensed
Histogram
improved formatting of displayed frequency distribution
BoxPlot
default for add.points option is overstrike instead of stack
corEFA
rotate="none" option replaces show.initial option, now deleted
Nest
the specification of the full model, the 3rd argument, can be
all the variables in the full model or now just the added
variables to the reduced model to define the full model
output control flags, in this order of presentation: explain, results, interpret
On by default, but each can be set within each procedure that generates a knitr
file as well as a global option, such as options(explain=FALSE)
---------------
Bug Fixes 3.3.1
---------------
library(lessR) added to knitr files from knitr.file option
SummaryStats
outliers properly identified if smaller in value than 3
Correlation
heat map for correlation matrix works more generally
Regression
works with no predictor variables, e.g., reg(Y ~ 1)
printed tables correctly display factor variables
SummaryStats
if a by variable, now no output for stats in assigned object, instead of
just for the last row
>>>>>> Changes for lessR version 3.3 (2015-03-19)} <<<<<<
-------------
Updates 3.3
-------------
knitr compatible
Regression, Histogram, SummaryStats
output system redesigned so that now all output is formally returned when
the corresponding function completes, back to the standard R way of doing
things in pieces, but here each piece is enhanced with additional features
new function regPlot, which produces the Regression plots from the saved
output of a previous regression run so that the plots can be interspersed
throughout a knitr document
new function print.outall which, to add to knitr functionality, allows each
of the pieces produced by Regression to be displayed individually, and is
called implicitly by simply entering the name of the object
new function print.outpiece which, to add to knitr functionality, allows each
of the paragraphs of output produced by Regression to be displayed individually,
including in knitr, simply by entering the name of the saved piece, such as
r$out_estimates if this was run: r <- Regression(Y ~ X1 + X2), and is
called implicitly by simply entering the name of the object
Regression
knitr.file option added to automatically generate knitr instructions which,
when processed, result in an enhanced html, pdf or Word document that
can be called interpretative output, statistical output plus commentary
graphics=FALSE option added, mostly for use with the new regPlot
explain=TRUE now generates the explanation in the knitr instruction file
instead of the console, and is now the default so reg.explain was removed
all new components for the saved object, now of class out_all from the
analysis instead of the object of class lm defined by the R lm function,
though many components are shared, also includes the knitr instructions
scatterplot matrix with correlations in the upper triangle now default
---------------
Bug Fixes 3.3
---------------
Screen size of subsequent plots not changed after Help()
SummaryStats
for a by analysis, levels with n=0 do not prevent analysis of all levels
Correlation
for a selected subset of variables, the heat matrix is plotted if requested
>>>>>> Changes for lessR version 3.2 (2015-02-24)} <<<<<<
-------------
Updates 3.2
-------------
if more than one plot is created from a function call the name of
each plot is displayed at the end of the console output
RStudio compatible, when in RStudio now graphics are managed as a
sequential stream to the plot window
---------------
Bug Fixes 3.2
---------------
ttest
line chart for a confidence interval of a mean displays (if requested)
even if no hypothesized value
for two groups, line.chart option works reliably for both groups
PieChart
color gradient for ordered colors from an ordered factor extended to all
color themes
col.ticks parameter no longer defined by lessR functions but passed
directly as an R parameter, which avoids the warning messages
warning messages
the causes of many warning messages, though benign, were identified and
removed by reprogramming
>>>>>> Changes for lessR version 3.1.1 (2014-09-22)} <<<<<<
-------------
Updates 3.1.1
-------------
corReorder
provides a new cor matrix if specified
BarChart
for gray scale, bars a little lighter shade of gray
corEFA
lavaan code from the EFA solution revised
label
no argument, label(), displays all variable names and labels
PieChart
passing standard R graphics parameters produces a square chart, so to
avoid this issue the magnification factors cex and cex.main, for
the labels and title, are explicitly defined
ScatterPlot
default 1-dimensional scatter plot is method="overplot"
ttest
graph for one group extends to large deviant values of mu0 from the data
brief version includes the margin of error
includes needed sample size for desired margin of error for 1 and 2 groups
Write
for a csv write, create a second file of any variable labels
---------------
Bug Fixes 3.1.1
---------------
label
if the specified variable does not exist, an error message is displayed
>>>>>> Changes for lessR version 3.1 (2014-02-24)} <<<<<<
-------------
Updates 3.1
-------------
default color theme
changed from "blue" to "dodgerblue", which now has 0.25 default
transparency for bar fill, if the previous "blue" is desired,
then set with: set(colors="blue")
citations
use of functions from other contributed packages cited in output
BarChart, BoxPlot, Histogram, ...
can specify an entire data frame for analysis with the data parameter
in addition to the variable parameter (x, usually listed first),
e.g., hs(attitude) or hs(data=attitude)
Density, LineChart
analysis of a data frame or list of multiple variables possible
BarChart
invisibly returns the frequencies and proportions just as SummaryStats
e.g., stats <- BarChart(Y), so stats contains this info
pre-set transparency level of col.fill.bar applies to bar chart bars
of a single variable
SummaryStats (and functions that call SummaryStats)
outliers listed in two groups, those above the high box plot whisker,
and those below the low box plot whisker, and if more than 25 then
the intermediate values in a group are not listed
more appropriate output when there is a frequency of zero on a
by variable
an explanatory note provided when computing row or column
proportions that result in divide by 0, which displays as NaN
for "not a number"
label
can assign a variable label to a variable as well as list the label, so
labels can be created/modified without reading from an external file
corCFA
lavaan code for the default maximum likelihood solution with the lavaan
function cfa generated for the specified measurement model
content of items by scale listed in the sorted order by loading
parameter added, labels="only" only lists the variable labels with
no analysis for a content analysis only
model solution invisibly returned that includes the estimated parameters
and the scale reliabilities plus residuals
improved formatting of column displays
corEFA
lavaan code generated for measurement model suggested by the EFA solution
simCLT
triangle package needed for antinormal distribution has been updated
to R 3.0, so antinormal distribution restored (antinormal distribution
has no values in the middle and most values at the extremes)
---------------
Bug Fixes 3.1
---------------
Merge, Recode, Subset, Transform
variable labels that exist in the input data frame(s) are retained
in the transformed data frame
BarChart, SummaryStats
for more than 10 categories the proportions are correctly computed
Correlation, ScatterPlot
correct variable labels listed
ScatterPlot
method parameter used for purpose other than specifying spearman or
kendall for correlation type, of which use is now flagged
corScree
specified correlation matrix analyzed instead of just one named mycor
corEFA
1 factor solution completes
>>>>>> Changes for lessR version 3.0 (2014-01-02)} <<<<<<
-------------
Updates 3.0
-------------
Subset
new parameter: random
Specifies the number or proportion of data rows to retain, which
replaces the dual use of the rows parameter to both perform
this task and provide a direct specification of the rows of the
data table to be included/excluded, so now the following work:
mydata <- Subset(c(1,4)) # retain only rows 1 and 4
mydata <- Subset(-c(1,4)) # delete only rows 1 and 4
ttest
new parameter: line.chart
When set to TRUE, adds a line chart of the response variable
for each group in the analysis
aesthetics of the density curve output updated
---------------
Bug Fixes 3.0
---------------
ANOVA
condition that lead to a warning for the means plot fixed
Correlation
in the output, the correlation matrix object was always described
as mycor regardless of the actual assigned name, this line of
output is now deleted
Density
missing data with specified bins now works
ttest
density plot in gray scale if colors="gray.black"
>>>>>> Changes for lessR version 2.9.7 (2013-10-29)} <<<<<<
-------------
Updates 2.9.7
-------------
details
add a brief version, details.brief, which only lists the table of
variable names and any variable labels
ScatterPlot
for 1-D scatter plot, to conform to standard R, rename option
plot.method to method
Read
relying upon the read.xls function from the gdata package, can
read Excel files identified by the .xls or .xlsx filetype,
both the data file and/or the labels file can be Excel files
add a brief version, rd.brief, which calls the new details.brief
provide an option to browse for the labels file, labels=""
always display the full path of the data file and any label file
---------------
Bug Fixes 2.9.7
---------------
BarChart
if a data frame analyzed, then a categorical variable with only
a single value would cause a fatal error, now the remaining
variables are analyzed and a diagnostic displayed instead
Read
labels files for Windows now properly specified
ScatterPlot
for a plot with a categorical x-axis, additional parameters such as
ylim now work correctly
corScree
graph of "successive differences of eigenvalues" now properly labeled
>>>>>> Changes for lessR version 2.9.4 (2013-08-25)} <<<<<<
-------------
Updates 2.9.4
-------------
BarChart
return the table of frequencies, so can assign to an object
BoxPlot
subset of variables can be specified, e.g., bx(c(x,y,z))
CountAll
parameters may now be added, such as bin.start for Histograms
Density
test for normality done only if a normal curve is plotted
Histogram
bin.end parameter added
subset of variables can be specified, e.g., hs(c(x,y,z))
ttest
to accommodate density plot of more data sets, bandwidth default
changed from nrd to bcv
two vector form of two-group t-test now accommodated from a data frame
to permit a dependent-groups analysis from a data frame
for a dependent-groups analysis, or paired t-test, a scatter plot of
the two variables is produced with a diagonal line through the plot
to indicate equality and the vertical distance from the line to each
point displayed to indicate the extent of the change
ttestPower
value of n on graph displayed as an integer or with decimal digits
as appropriate
ScatterPlot
for a scatter plot of two numeric variables, diag=TRUE places a diagonal
line through the plot with vertical lines from each point to the diagonal,
primarily for plotting change in a dependent samples t-test
removed x.start, x.end, y.start and y.end for bubble plots: use xlim, ylim
SummaryStats
returns summary statistics for analysis of a single variable
subset of variables can be specified, e.g., ss(c(x,y,z))
analysis of a data frame yields the default value of brief, which
can be overridden in the function call
---------------
Bug Fixes 2.9.4
---------------
BarChart
graceful termination if a bar chart is attempted with only 1 unique value
Density
col.fill.nrm, normal curve fill, can set to transparent for blue color theme
Regression
density lines in residuals density plot now appropriate color for
black backgrounds
ScatterPlot
xlim and ylim also applies to bubble plots, before they were ignored
line plot by default even when intervals of successive values of a
sorted x are only equal to within 9 decimal digits
1-D plot displays outliers with same plot.method as regular points
Correlation
name of first variable in bivariate correlation now displays correctly
>>>>>> Changes for lessR version 2.9.3 (2013-05-26)} <<<<<<
Note: The Excel read functionality added in 2.9.2 is removed because it required
Java, and this additional installation was adding too much complexity
for users. To retain this functionality, do the following.
> install.packages("xlsx") # one time only
> library(xlsx) # for each R session to invoke the following
> mydata <- read.xlsx(file.choose(), sheetIndex=1)
This provides for a direct read of an Excel file by browsing for the file.
To specify a specific path name or URL, replace file.choose() with the
correct name in quotes.
The only lost functionality if the above code is implemented is that variable
labels cannot be read with an Excel file. To provide for these labels first
save the Excel file as a csv file.
-------------
Updates 2.9.3
-------------
LineChart
a "zero" option is provided for center.line to pass the line through 0
simCLT
the "antinormal" option is inactivated until the supporting triangle
package is updated
---------------
Bug Fixes 2.9.3
---------------
BarChart
does not terminate when a table is specified as input
>>>>>> Changes for lessR version 2.9.2 (2013-05-08)} <<<<<<
-------------
Updates 2.9.2
-------------
Read
tab-delimited text data file detected by default in addition to
csv text data file
Excel files now read and detected by default, including variable labels
Density
for colors with a black background, density functions plotted with
light colors
corRead
abbreviation rad.cor no longer available, use rd.cor
---------------
Bug Fixes 2.9.2
---------------
Histogram
rounding error in the computation of cumulative probabilities fixed
>>>>>> Changes for lessR version 2.9 (2013-03-11)} <<<<<<
-------------------
New Functions 2.9
-------------------
Nest
compare a nested model to a full model with least-squares or logit fit
details
obtain the details of a data frame, such as called from Read
-------------
Updates 2.9
-------------
a variable to be analyzed from the user's workspace is so noted
Read2
renamed from rd2
Regression
can return an object of class lm
Logit
classification table added
if only some forecasts shown, the middle range is for fitted values
close to the threshold of 0.5
collinearity analysis added for multiple predictor variables
can return an object of class glm
ANOVA
ANOVA tables now cleanly formatted
residuals displayed as in Regression, by default first 20 sorted
res.rows and res.sort options added, as in Regression
Histogram
can return an object of class histogram
Density
can return an object of class density
BoxPlot
allow R graphics parameters to be passed, such as whiskcol, see ?bxp
colors adjusted for gray and gray.black
can return an object with standard boxplot components
SummaryStats
if integers in input data then output to 2 decimal digits
if more than 50 outliers, then just first and last 25 are displayed
set
color white added
quiet option now can be set, e.g., set(quiet=TRUE)
brief option now can be set, e.g., set(brief=TRUE)
---------------
Bug Fixes 2.9
---------------
ANOVA
brief form works correctly
Logit
plot of fitted values and scatter plot produced consistently
BoxPlot
numerical values on the correct axis for vertical and horiz orientation
Histogram
situation in which largest value exceeded the largest bin fixed
set
transparency properly initialized for default blue
>>>>>> Changes for lessR version 2.8 (2013-02-01)} <<<<<<
The keepers of CRAN have changed the rules. They no longer
allow a function to automatically direct output to a data table.
They have decided that users should always explicitly specify
the destination of the output file.
What that means for lessR is that any function that outputs a
data table now cannot automatically write that data table to
mydata or other chosen name. Instead you must now explicitly
assign the output data table name when reading or modifying data,
usually use mydata or mycor. To do this, use the R assign
notation, <- , which assigns anything on the right side to whatever
is on the left side of the expression.
> mydata <- Read()
> mydata <- Transform(Y=X/12) also Subset, Merge, Recode, Sort
> mycor <- Correlation()
If you do not make this explicit assignment, the function still
works, but the output is dumped at the console instead of sent
to a stored data frame such as mydata to be available for later analysis.
Other generic changes:
To be consistent with R functions, the data frame option has been changed from dframe to data. Usually this is not used as the default mydata is relied upon, but now specify other names with the data option.
Previously the brief option was used inconsistently. For some functions it lead to a brief output, and for others it suppressed output. It remains for functions that primarily send output to the console. For graphics functions and data modification functions, now completely suppress output with quiet=TRUE.
-------------------
New Functions 2.8
-------------------
Merge
merges two data frames either horizontally or vertically
-------------
Updates 2.8
-------------
default system setting
n.cat, the maximum number of unique values of a variable to be treated as
a categorical variable by default, changed to 0, turned off by default
Recode, Sort, Subset, Transform
now precede the function call with, for example: mydata <-
Read
now precede the function call with, for example: mydata <-
variable labels now incorporated directly into the data frame and
are now read with the labels option
rd is the abbreviation, though the older rad is still available
lessR.data option re-specified as format="lessR"
quiet option replaces brief
Subset
holdout sample can no longer be created from with the function given the
rule change from CRAN, but holdout=TRUE creates the code to copy and
paste back into R to create the holdout sample
ttest
for two group analysis from a formula, the separate data vectors
are returned for later analysis (see the examples)
graph for two group analysis now in gray scale for colors="gray"
when input is summary stats, reported summary stats are to the
same level of precision as to what was input
variable label, if present, appears on density graph
standard R alternative option available for one-tailed tests
paired=TRUE option available for dependent-groups t-test
ttestPower
powercurve.t.test name removed in favor of ttestPower
ANOVA
randomized blocks analysis displays the marginal and grand means
two-way between groups analysis displays the cells size once instead of
the same number for all cells
Regression
standardization option available
rgl package bug apparently fixed, so scatter.3d=TRUE is again available
for models with two predictor variables
singularity check added and solution terminated if so
residuals vs fitted Values plot plotted with current color theme
scatterplot of prediction intervals with current color theme
Correlation
can use method="kendall" and method="spearman"
graphics=TRUE to create a scatter plot matrix and heat map
pdf=TRUE to create and write scatter plot matrix and heat map to pdf files
Density
summary statistics reported by default
quiet=FALSE replaces text.out=TRUE
Histogram
quiet=FALSE replaces text.out=TRUE
ScatterPlot
one variable, method="jitter" option from R stripchart possible
missing data removed to enable ellipse from car package
the by variable need no longer be a factor
fit.line applies to each level of a by variable
for a by variable, width of plot adjusted for legend when saving to a pdf
quiet=FALSE replaces text.out=TRUE
BoxPlot
color of the box more vivid
quiet=FALSE replaces text.out=TRUE
Logistic
abbreviation is lr instead of older lgt
set
default colors="blue" bar fill is lightsteelblue3 from lightsteelblue
colors="sienna" and "gray.black" color themes added
colors="dodgerblue" given a light gray background
BarChart
reports the corresponding chi-square test
count.names option name changed to the more meaningful count.levels
quiet=FALSE replaces text.out=TRUE
PieChart
for consistency with other functions, col.pieces changed to col.fill,
the specified color of the regions of the pie chart
quiet=FALSE replaces text.out=TRUE
SummaryStats
only report summary statistics (chi-square test moved to BarChart)
---------------
Bug Fixes 2.8
---------------
ANOVA
pdf=TRUE properly writes the graphs to the working directory
ttest
graph for two group analysis shows the degrees of freedom in the title
if missing a grouping variable data value, analysis still proceeds
BarChart
ordered progressions of color with purple, sienna and dodgerblue work
Histogram
col.ticks warning addressed and no longer generated
text.out can now be set to FALSE
Density
a perfectly symmetrical distribution properly plots as a density function
LineChart
default area under the plotted line segments now fills to proper
color according to the current color theme
ScatterPlot
transparency for one and two variable plots correctly provided by default
ellipse works correctly for non-regular plots such as bubble plots
for kind option, bubble and sunflower can be now specified as documented
sunflower plot has background and grid colors according to color theme
show.n=TRUE works correctly for pairwise deletion for correlation matrix
Correlation
properly accept variables in global environment
Regression
prints residuals and forecasting errors when there is a factor predictor
for categorical variables the results for all levels are displayed
set
n.cat no longer set to 4 when set function called
Write
if suffix .csv or .rda already exists, not added again to file name
>>>>>> Changes for lessR version 2.6 (2012-10-24)} <<<<<<
-------------
Updates 2.6
-------------
Graphics procedures
Color themes were enhanced and the terminology for modifying individual
colors in a specific graph or system-wide with the set function was
standardized. 'fill' refers to the color of an interior region,
either of a bar or a circle. 'stroke' refers to a line or outline,
such as the border of a histogram bar or a plotted point. Also, to
change a color theme is now only available with function set, as
are references to transparent colors with trans.fill.bar and
trans.fill.pt.
set
added an orange color theme, which has a black background instead
of the usual light background, also added dodgerblue and purple
revised green color theme
added option ghost to provide transparent bars against a black background
with no grid lines, which works well with colors such as orange and red
colors option was getting too complex and was cluttering the options lists
for the graphics functions, so color theme and transparency only
available from the set function, but with more extensive options
Read
default text output to console that describes the data is redesigned
new parameter lessRdata allows direct reading of built-in data sets
can read directly from built-in data sets with lessR.data= option
Write
write any specified data frame, not just the default mydata
specify any file name or rely upon the default
by default write row IDs as part of the written csv file
the dframe option moved to the end of the parameter list to be
consistent with Read
Subset
abbreviation locate added to emphasize locating cases without creating
a new data frame, where save.dframe is automatically set to FALSE
can subset on row.names
criterion for selecting rows, rows, can be an integer or proportion,
to indicate the number of rows to randomly extract and also to create
a hold out sample
Recode
just data for the variables to be recoded are shown before the recode,
and just the recoded and any new variables shown after the recode
a recode is not allowed to be applied to a factor because doing so
converts the factor to character strings (use factor function instead)
Transform
transformed data is shown only for variables that have been transformed
Sort
default saves the sorted data frame written over the input data frame
without needing to explicitly assign the result to a data frame
the keep=FALSE option allows the sorted data frame to be written to
another data frame with the R assignment statement
random option added to randomly shuffle the rows of data
Help
the argument for a specific help page no longer needs to be enclosed
in quotes
capitalization of the argument for a specific help page is irrelevant
ANOVA
randomized block design supported (in addition to one and two factor
between groups designs)
fitted plot and data plotted for randomized block design
residuals provided
effect sizes provided
graphs may be saved to pdf files with pdf=TRUE
one-way cell mean plot works with current color theme
HSD analysis for two-way models, between groups and randomized blocks
marginal means provided for two-way models
Regression
scatterplot matrix incorporates the color theme
display of prediction intervals includes interval width
decimal digits uniformly applied across the text output
Logit
scatterplot matrix added when there are multiple predictor variables
ttest
option for saving the graphic of the two density curves to a pdf,
consistent with other lessR functions for graphics
add a show.title option to suppress the title over the graph
Cohen's d effect size index added to one-group t-test
density plot with Cohen's d, mean and hypothesized mean
ScatterPlot
for one variable, dot plot, gray scale outliers displayed in squares
and diamonds, for potential and actual outliers, respectively
Histogram
trans.bars option available, analogous to trans.pts for scatter plots
LineChart
option col.border added to specify the border color of the filled
polygon under the plotted lines, including the value of "transparent"
BoxPlot
dotplot option changed to add.points to be consistent with the call
to the ScatterPlot function for one variable, i.e., a 1-D scatter plot
corCFA
sum of squares and average residual for each item and total available
the number of default iterations for communality estimates increased
from 15 to 25
an abbreviation called scales added to retain 1's in the diagonal for
component analysis, that is, the observed scale scores
corEFA
items by default sorted by their highest factor loading, with an
option provided to not do this
min.load option changed to min.loading and this applies to the
output of the EFA as well as the constructed model for the CFA
n.fact argument changed to n.factors
data files
the name of each included data file begins with "data" instead of "dat"
prob.tcut
renamed from qnt.t, t-cutoff probability function with t and normal curves
---------------
Bug Fixes 2.6
---------------
Subset, Transform, Sort
when dframe not saved because save.dframe set to FALSE, dframe
properly is assigned to a new data frame via an assignment statement
PieChart
colors option works correctly
BarChart
changes to background color and grid with colors="gray" work correctly
frequency table displayed when prop=TRUE
LineChart
color theme now applied to fill color under plotted polygon
ttest
standard deviation on graph for second group reported correctly
>>>>>> Changes for lessR version 2.5 (2012-08-09)} <<<<<<
-------------------
New Functions 2.5
-------------------
Sort
sorts the rows of a data frame by the values of specified variables for
both numeric variables as well as factors
Transform
a modified version of the standard R transform function, but by default
saves the revised data frame to the input data frame and provides
feedback and information regarding the transformation(s)
Subset
a modified version of the standard R subset function, but by default
saves the revised data frame to the input data frame and provides
feedback and information regarding the changes to the data frame
Logit
logit analysis, a wrapper for the standard R glm function with
family=binomial plus related functions such as summary and predict
-------------
Updates 2.5
-------------
corCFA, corReflect, corReorder
variables are now specified by their names instead of by their ordinal
position in the correlation matrix
corEFA
to match the change in specifying variables in corCFA, the derived
confirmatory model is now written in terms of variable names
corScree
on the graph of the differences of successive eigenvalues, a horizontal
line is drawn to better highlight the "scree"
Correlation
now can provide a list of variables from the input data frame instead
of having to first separately create the subset data frame
now non-numeric variables are now automatically deleted from a submitted
data frame or variable list with the analysis proceeding
Recode
a list of variables instead of just one variable may now be recoded
missing data entries may now be recoded to valid values
specified valid values may now be recoded to missing values
ScatterPlot
original function Plot, abbreviated plt, was based on R function plot,
which did a scatter plot of two variables and also also did a line chart
for one variable, now only 1 and 2 dimensional scatter plots are done,
so function renamed accordingly, where a 1-D scatter plot is a dot plot,
though Plot is still available as a name
Help
each help page can now be invoked with a variety of key words, which
usually include the full and abbreviated names of each function
described on that help page
---------------
Bug Fixes 2.5
---------------
SummaryStats
now recognizes n.cat to treat numeric variables as categorical if the
number of unique values is less than or equal to n.cat
corCFA
when default sort option on, sometimes items were not sorted properly
set
the transparency level of plotted points in ScatterPlot now works
>>>>>> Changes for lessR version 2.4 (2012-07-21)} <<<<<<
-------------------
New Functions 2.4
-------------------
Recode
recode individual values of an integer or factor variable
The following new functions work with a correlation matrix, named
mycor by default, instead of the data matrix from which the correlations
are computed. Each function that outputs correlations also generates a
heat map of the output matrix.
corCFA
confirmatory factor analysis and item analysis for multiple indicator
measurement models from an input correlation matrix
corEFA
exploratory factor analysis based on R factanal function, though also
provides for a multiple indicator measurement model based on the
exploratory analysis and the corCFA code for which to analyze the model
corList
list the ordinal position of each variable in the input correlation matrix
to facilitate using the other correlational routines
corProp
calculate proportionality coefficients from an input correlation matrix,
used to identify items that are indicators of the same factor
corRead
read an input correlation, or other square, matrix
corReflect
reflect specified variables in an input correlation matrix
corReorder
re-order the specified variables in the input correlation matrix
corScree
eigenvalue plot and plot of differences of successive eigenvalues to help
determine the number of factors
-------------
Updates 2.4
-------------
System wide
Variable labels when applied to axis labels on a graph are now truncated
to 50 characters for y-axis and 45 characters for x-axis to fit
All graphic files can now be saved from the call to the graphic function
as preceding the function call with an R pdf statement does not work
due to the customized graphics system that allows the Help window
to persist across analyses
Cutoff value to interpret a numeric variable as categorical now called
n.cat instead of n.cut, and is implemented system wide with the
set function
Correlation
minimum default number of digits in output correlation matrix is 2
computed correlation matrix automatically written to mycor
missing data choices made explicit with parameter miss, pairwise is default
cell-wise sample size reported for pairwise deletion
effective sample size for all cells reported for listwise deletion
heat map added when a correlation matrix is computed
Plot
provide for a by variable, a grouping variable, for which the points
are plotted in a different color and/or shape for each value of by
SummaryStats
IQR added to full version
BarChart
legend placement option for right margin added to the standard R legend
locations and is now the default, legend.loc="right.margin"
legend in right margin accommodates variable labels
for displayed cross-tabulation table, variable names instead of
variable labels used
for count.names option, no longer needed to place the data frame name
and a $ in front of the specified variable name
text.out option added so can be set to FALSE
when applied to a data frame, individual graphs written to individual files
Histogram
text.out option added so can be set to FALSE
when applied to a data frame, individual graphs written to individual files
BoxPlot
text.out option added so can be set to FALSE
when applied to a data frame, individual graphs written to individual files
ttest
graph option added, if FALSE then no graph is produced for two groups
LineChart
changed name from RunChart to better reflect its more general meaning
Deprecated function names removed
color.barchart, color.boxplot, color.density, color.hist
Deprecated function names renamed
sim.CLT to simCLT, sim.CImean to simCImean,
sim.flips to simFlips, sim.means to simMeans
---------------
Bug Fixes 2.4
---------------
Summary Statistics
labels for the two variables in a cross-tab no longer switched
to
remove debug print statement
ttest
extra null graphic window no longer generated for two groups analysis
BarChart
left margin on horizontal bar chart was sometimes too large
color theme for a single variable now properly displays
>>>>>> Changes for lessR version 2.3 (2012-06-10)} <<<<<<
------------
Overview 2.3
------------
Naming Conventions
------------------
All data analysis functions now have two names, a longer, more descriptive name, that
involves uppercase letters, such as Summary Stats, and a short abbreviation, here ss.
Either version is equivalent. The purpose of the uppercase letters is to distinguish
lessR functions from the standard R functions with similar names. When appropriate,
functions can also have an abbreviation such as brief to indicate a briefer form of
output, here ss.brief.
Color Themes
------------
Using the new set function, the colors options sets the system wide color theme.
The default is "blue" and several other colors are available, including gray scale
with "gray". The colors option may also applied to any one specific graphic
function call to set the color theme just for that one resulting graph.
Transparency for plotting individual points with Plot is also available with the
trans.pts option, from 0 to 1, with a 0 being opaque and a 1 being fully
transparent. The trans.pts option may be set with the set function for all
subsequent analyses, or it may be set for any one specific call to Plot.
Direct Reference of Variable Names (from 2.2)
----------------------------------
lessR defaults the name of the data frame with data read from the Read or rad
function with mydata. Now this convention is leveraged by dropping the need for the
R attach function, or by having to include the data frame name and a $ in front of
the variable name, or using the with function. Instead, for each specified variable
name, lessR searches the user's workspace, the global environment, as well as the
data frame mydata, or the specified data frame name, for the relevant variable.
Variable Labels (from 2.2)
---------------
lessR functions now can access variable labels, which will replace the variable
names on the axis labels for graphic output, and be displayed adjacent to the
variable names on text output. Use the labels function to access the variable
labels for standard R functions. See help(Read) for directions on how to enter
the variable labels.
Improved Appearance of Graphs
-----------------------------
In addition to the new color themes, the appearance of the graphs has been changed
to print the values along each scale of an axis is a smaller font and a shade of
gray instead of black.
-----------------
New Functions 2.3
-----------------
Model, model
Function for a linear analysis, which automatically calls the relevant function --
ttest, ANOVA or Regression -- and therefore replaces those functions from the
user's perspective.
set
Created as a wrapper for options(colors="xxx") added for graphics routines to
specify a color theme.
to
Created to generate variable name lists with sequential numbers when reading
data into R such as from a csv data file.
-----------
Updates 2.3
-----------
Help, hlp
updated help window opens when lessR is loaded
Read, rad
Add option for reading native R data files
SuppressWarnings to avoid warning message on read.csv for Excel csv files with no
last
SuppressWarnings on SPSS files to avoid "Unrecognized record type 7"
Write, wrt
Add option for writing native R data files
Automatically add a file type, either .csv or .rda
ttest, tt
Also do analysis of not assuming equal variances.
Set extra decimal digit for analysis from stats (already set from data)
By default, at least two decimal digits
Added consistent formatting to numerical output according to digits.d
Two-group density graph has density scaling removed from y-axis
Two-group density graph has smaller font sizes and margin adjustment
BarChart, bc
Brief stat output now the default, bc.brief removed
Add error condition of not having col.bars and colors both activated
Add colors="gray" option and bc.gray
Display tick marks and tick labels in dark gray
Histogram, hst
Add vertical grid in addition to existing horizontal grid
Add colors="gray" option and hst.gray
Display tick marks and tick labels in dark gray
DotPlot, dp
Add colors="gray" option and dots.gray
Display tick marks and tick labels in dark gray
Density, dens
Remove horizontal grid, leaving no grid
Add colors="gray" option and dens.gray
Display tick marks and tick labels in dark gray
Plot, plt
Add colors="gray" option and plt.gray expression
Display tick marks and tick labels in dark gray
Add ncut to treat x as a factor if too few unique data values
When x is a factor, now do summary stats of y by each level of x
Run chart now lists n and n.missing
Scatterplot matrix and correlation matrix added for a data frame
Regression, reg
3D scatterplot optional for two predictor variables instead of required
The colors setting applies to reg graphs
BoxPlot, bx
Consistent formatting of text output with default decimal digits
Correlation, cr
Correlation routine pulled from plt and made its own function
The version cr.brief added
Correlation matrix of a data frame added
RunChart, rc
old access was Plot with one variable, which now produces a dot plot
prob.norm
now returns the probability in the console like R pnorm function
qnt.t
renamed from prob.t as it is the quantile that is returned
now returns the quantile in the console like R qt function
-------------
Bug Fixes 2.3
-------------
system wide
a variable named with the name of an R function is now permissible
ttest, tt
Analysis from summary statistics does not need a mydata data frame
to exist prior to the function call
Variable labels no longer switched on response and grouping variables
Labels now work with one group analysis
Properly align group1 and group2 output to the user's workspace
with the correct group
Plot, plt
Run plot failed if missing data
Plot with factor on x-axis failed plot of means if missing data
If missing data, do not try ellipse which fails
For bubble plot, x-axis now has proper scale
Summary Statistics, ss
The null graphic window no longer opened
BoxPlot, bx
When doing a dot plot, colors did not transfer from bx
>>>>>> Changes for lessR version 2.2 (2012-03-28)} <<<<<<
-----------------
New Functions 2.2
-----------------
sim.CLT
Simulation for Central Limit Theorem
sim.flips
Simulation of coin flips
sim.CImean
Simulation of confidence interval
sim.means
Simulation of repeated sampling of means
prob.t
Probabilities of t-distribution
-----------
Updates 2.2
-----------
Note: The following long form names are not valid as function calls until Version 2.3.
graphics routines
Font for scale values along each axis smaller and in a dark gray
ttest, tt
Rewrite to allow one or two groups, data or summary stats
Allow missing data
Restore graphic parameters so top margin of graphics window not too large
tt.brief option added
BarChart, bc
Count.names option for reading counts directly from a data file with counts
Smaller font for legend and no legend border to better display
Smaller font for axis values
For horizontal plots, horizontal labels and accommodate space in left margin
bc.brief added, brief=FALSE is new default for bc
Histogram, hst
Smaller font for scale
Plot, plt
Smaller font for legend and no legend border for 2 variables to better display
Provide covariance coefficient
Summary Statistics, ss
ss.brief added
Correlation, cr
Provide covariance coefficient
Density, dens [now den in 2.3]
Provide densities on vertical axis as an option instead of a requirement
Read, rad
Read spss (.sav) files in addition to csv data files
For rad.both, have labels display correctly
prob.norm [normal curve probabilities]
Only give normal densities on vertical axis as an option
Add second x axis, z-scores
Make vertical, density, axis as an option
Scale axis labels to .9, add mag option
prob.znorm [normal curve display with z-scores]
Default y-axis to null, add as an option
Add z-values as a default
Scale x-axis according to standard deviations
Scale axis labels to .9, add mag (magnify) option
stats.t.test
Removed, incorporated into ttest which now processes data from
summary statistics or the data
-------------
Bug Fixes 2.2
-------------
Regression, reg
If residual is 0, Nan's lead to missing data no longer causes
Cook's distance function, and therefore the entire function, to fail
>>>>>> Changes for lessR version 2.1 (2012-02-08)} <<<<<<
-----------------
New Functions 2.1
-----------------
dots [in 2.3 rename to DotPlot, dp]
created the function
pieplot [in 2.3 rename to PieChart, pc]
created the function
-----------
Updates 2.1
-----------
package
add citation
all lessR functions that read data have attach requirement removed
relevant lessR functions have automatic use of variable labels
variable labels function label
rad [in 2.3 also named Read]
display name of file read at the beginning of the output only if rad()
default is now to not attach mydata
add read labels options with rad.labels and rad.both
add max.lines options and display full data/labels when applicable
convert display option to brief option, add function
reg [in 2.3 also named Regression]
add error check for no data frame, which is required
for Background, specify number of obs retained for analysis
add references
add variable labels to Background section where variables are listed
add reg.brief and reg.explain methods
reformat Basic Analysis output to print all values individually
if a non-numeric variable in model, then do not attempt scatterplot matrix
if a non-numeric variable in model, then no scatter.3d plot
plt [in 2.3 also named Plot]
in title, use actual variable names instead of "x and y"
loess fit line replaces lowess, along with access to loess span parameter
use dates from an existing time series
add missing data count
put error traps for calling with the wrong data types
barchart [in 2.3 rename to BarChart, bc]
reformat output to much more compact
for UseMethod, evaluate class of 1st attribute only to avoid a warning
add a y-axis label
get border option to work and change name to col.border
properly switch axis labels if horiz=TRUE
add warning message for beside option off with only a single variable
add warning message if addtop set for a horizontal bar graph
make chisq the default, and reformat output
put the variable names on the tables for row and column proportions
re-scale bar width for 2, 3 or 4 bars, from 1 var or stacked 2 var
get vivid option to work for gradient applied to ordinal data
for vertical graphs of two vars, make legend horizontal with addtop room
for data frame, numeric data types of few unique values treated as categorical
for data frame, add dev.off() when finished with graphs
enhanced the color palettes with R palettes of rainbow, heat and terrain
make stacked chart for two variables the default
smooth [in 2.3 rename to Density, den]
add bw parameter
make no density axis the default, but add y.axis to include if desired
add normality test
allow for missing data
change col.hist to col.bars
get color of the plotted curves working
histogram [in 2.3 rename to Histogram, hst]
move Number of Bins output next to freq dist, add Bin Width
summary statistics and label now with describe.numeric
scientific notation turned off for histogram plot
col and border options renamed to col.bars and col.border
summary stats now provided
for data frame, numeric data types of few unique values treated as categorical
for error message regarding bin range, turn off scientific notation
boxp [in 2.3 rename to BoxPlot, bp]
add values of outliers to text output
adjust axis labels for vertical box plot
provide for default colored background and grid lines
describe [in 2.3 rename to SummaryStats, ss]
formula input changed to by= option
output reformatted to much more compact, and extended
outlier detection added to description of numerical variables
if too many values, then just report counts
if all values unique, just report the values and a note as an ID field
for numeric, if digits.d > 10, output size changed to 4 with prompt to override
for data frame, numeric data types of few unique values treated as categorical
add skewness, kurtosis
add a brief=TRUE option, which works for both numeric and categorical variables
smd.t.test [in 2.3 rename ttest,tt]
add brief option and function
stats.t.test [in 2.3 incorporated into ttest]
add one-sample option
by default resolve number of digits from precision of entered stats
help.me [in 2.3 now named Help, hlp]
update color.hist [in 2.3 Histogram, hst] description
-------------
Bug Fixes 2.1
-------------
plt [in 2.3 also named Plot]
missing data caused an error in bubble plot
reg [in 2.3 also named Regression]
allow missing data