NEWS

ca 0.71 (2018-10-10)

Fixed issue wirth subset MCA (computing row coordinates gave an error 'Error in (Z/Q) %*% col.sc [...]')
Added row-and columnnames for returned matrices (indicator, Burt, 'subinertia') in MCA
Fixed a (rare) issue with JCA and a subset not containing columns from all original factors

mjca() with 'supcol!=NA': - Fixed Mass, ChiDist and Inertia values in print method. - Fixed contributions in summary method.
mjca() with 'subsetcol!=NA' and 'lambda=JCA': - Fixed warnings message ('In rep(1:Q.sub ...)')
Fixed row output for 'lambda!="indicator"' in mjca (rows were previously referring to the Burt matrix, now rows are from the original data)
Added cacoord() for extracting CA/MCA coordinates (standard/principal coordinates, and row-/column configurations for plotting)
Added caconv() for converting between CA/MCA data types (frequency tables, indicator/response pattern/Burt matrices)
mjca() has been made an S3 generic, with data.frame, table, and array methods
added .arrows() to draw nicer arrows with arrows=TRUE in plot.ca() and plot.mjca()
plot.ca() gains a lwd= argument for arrows and lines
Updated mjca.Rd with details of lambda= argument
Added multilines() for nicer mjca plots
plot.mcja gains a collabels= argument determining the form of column labels

mjca(): Added option 'reti' (logical, if TRUE returns the indicator matrix as $indmat).
mjca(): Fixed potentially negative eigenvalues (numerical issue; negative ev's are set to zero).
mjca(): Burt matrices ($Burt) now with row- and columnnames.
print.summary.ca(): Fixed scree plot.

Fixed bug in print.summary.mcja re columns
In plot.ca() and plot.mjca(), for labels=(1,1), don't offset the labels from the points

Begin to assign dimnames to matrices from ca() and mjca()— for now, just the coordinates matrices
mjca() now returns an additional component, factors, containing factor names and levels, to facilitate plotting and manipulation. This can be used to customize the color, symbols, and labels in plots.
summary.ca() gains rows= and columns= arguments that can be used like scree= to suppress parts of the output.
summary.mjca() gains a similar columns= argument
Fixed calls to rgl functions in plot3d.ca.r to avoid 'no global binding' warnings.
Fixed bug in plot.mcja() with dimension percentages for lambda="adjusted"

plot.ca() and plot.mjca() now provide sensible default xlab and ylab
mjca() now accepts a table object as input, using internal expand.dft
ca() has been made generic, accepting input in the form of matrices, data frames (coerced to frequency tables), objects of class "xtabs" or "table" and one-sided formulae of the form ~ F1 + F2, where F1 and F2 are factors.

Added xlab=, ylab= and lines= arguments to plot.ca(). The function now returns an invisible result containing the row and column coordinates calculated internally for the plot, for use in additional plot annotation.
Added xlab=, ylab= arguments to plot.mjca(). The function now returns an invisible result containing the row and column coordinates calculated internally for the plot, for use in additional plot annotation.
For plot.ca(), the defaults for xlab= and ylab= were changed from "" to "_auto_", meaning that the axis labels are auto-generated to be of the form "Dimension X (xx.xx%)" in conformance with other dimension reduction plots.
Added col.lab argument to plot.ca(), to provide colors for the point labels, similar to col for points.