| NEWS | R Documentation |
Minor correction to reflect change in behavior of '&' operator with new release of R.
A substantial revision to this package in still planned to occur with version 0.6.0.
Removing plot method for Surv objects, so as to
avoid overwriting the new plot.Surv method in
package:survival.
A substantial revision to this package in planned in the months ahead and is planned to occur with version 0.6.0.
Minor update to take account of the release of package:ggplot2 2.2.0.
Corrected error in sf. Thanks to Walton Jones for
bringing this to my attention.
Added some xtable methods applicable to survival data.
Corrected error in comp when comparing two groups. This was
present in Versions 0.5.0 and 0.5.1. Thanks to onesandzeroes and drcanak for pointing this out.
Also re-named the vignette again.
Have included profLik function from version 0.4.6
again.
Re-named the vignette.
This is a re-write 'from the ground up'. A more 'oject-oriented' approach has been adopted. The 'ten' function/ object has replaced the former 'tne' method. (Motivated in part by the user who commented that "I'm afraid the code in the function '.getTne' appears somewhat cobbled together".)
A number of functions and datasets have been left out temporarily for this release in order to focus on the 'core' package elements. These will be added back in due course, most likely version 0.6.0.
cutpError in calculation of log-rank test in some in
cases has been fixed. Code also simplified to make better use of data.table.
Thanks in particular to Rached Alkallas for her helpful
work on this. As suggested:
There was an problem with the previous version [0.4.6], which
may have affected the scientific results of those using this
function
compProblems when calculating the 'test for trend'
have been fixed. Thanks to Karl Ove Hufthammer for pointing this
out. I have also modified the help for comp which was using
the words 'groups' and 'strata' interchangably. This has been
corrected elsewhere in the manual. comp does *not* yet work
with stratified models, but this will be implemented in due course.
Thanks also to Mario Ouwens for input on the Renyi function and
Haitao Yang for comments on the log-rank test.
autoplot.survfitFunction has been improved. This will
likely be deprecated in due course in favor of autoplot.ten.
The returned object will now print using
autoplot.tableAndPlot.
Examples of changing the output are now given in the 'plots'
vignette. These include y axis limits and the order of the
legend labels and position. Thanks to Line Heylen for these suggestions.
Thanks also to Jason Kennedy for bringing up the problem of legend
re-ordering.
rsq.coxphTypo fixed. Thanks to Nir Y Krakauer for bringing this to my attention.
Error in use of dQuote in comp now fixed.
Thanks to all who pointed this out: J Smith, M Fiocco, D Winsemius,
J Bienkowska, A Calcareo, N Villanueva, A Lover, R Perry, K
Karakostas.
Typo in autoplot.survfit fixed. Thanks to D Menne.
Example in autoplot.survfit now includes method to customize the
output further. Thanks to S Melov and L Neves.
Scoping in comp fixed. Thanks to A. Atkinson.
8/14/14: Add cf6.c to src; use this instead of depending on survival package.
7/6/2014: Add gof method.
7/2/2014: Fix .getTne to work on Win and Linux.
6/21/2014: Change name of autoplot and push new version
to github.
6/9/2014: Improve autoplot.survfit; coloring now
consistent for line and fill.
Add ci method.
Move documentation from quantile to ci.
5/21/14: Fixed autoplot.survfit: arbitrary alpha now
works with upper box in filled survival lines.
6/6/2014: Fixed .getTne, was not calculating values correctly;
comp() now works correctly.
5/20/14: Fixed comp, compNSurv for cases where
last element in
covMatSurv is NaN.
Fixed sig so likelihood-ratio test works for factors
(needs call to model.matrix).
Fixed cutp - need to convert data.table to
data.frame in function body for now.