RELEASE HISTORY OF THE "care" PACKAGE


Version 1.1.11:

- updated URLs.


Version 1.1.10:

- slm() now also reports the unexplained residual error.
- predict.slm() now reports the standard deviation
  of the predictive distribution (attached as attribute "sd").
- added DOIs for all references.


Version 1.1.9:

- added import statements required by R-devel.


Version 1.1.8:

- fix "S3 generic/method consistency" NOTE raised by R-devel.


Version 1.1.7:

- slm() and slm.models() have been merged into a single function,
  with slm.models() removed.  Correspondingly, the "slm" object 
  returned by slm() now may contain several models.
- predict.slm() has been changed accordingly. 
- slm() now also works if there are no predictors.


Version 1.1.6:

- the example R scripts (brain and diabetes data) are now provided in 
  R notebook format.
- change of maintainer email address.


Version 1.1.5:

- change of maintainer email address.
- corrections to index.html file in inst/doc folder.


Version 1.1.4:

- added import statement in NAMESPACE file to address warnings in R
  version 3.1.0.
- added example scripts for Efron et al. (2004) diabetes data and 
  Lu et al. (2004) gene expression data.
- now suggests "crossval" package for estimating prediction accuracy.


Version 1.1.3:

- small changes to get rid of warnings in R-devel
- update of dependencies


Version 1.1.2:

- NAMESPACE file added.
- now requires corpcor version 1.6.2.


Version 1.1.1:

- reference to Zuber and Strimmer (2011) added.
- lambda and lambda.var options instead of shrink=TRUE/FALSE
  to follow the conventions in the corpcor package.
- car.models() renamed to slm.models()
- new function make.predlist().


Version 1.1.0:

- depends now on R 2.10.0 and corpcor version 1.6.0
- much faster computation of CAR scores in function carscore()
  due to the use of crossprod.powcor.shrink() in corpcor package
- syntax of carscore() function now the same as the catscore() 
  function from the sda package
- fitlm() renamed to slm()
- slm() is now also much faster due to the use of the 
  crossprod.powcor.shrink() from the corpcor packages 
- new predict.slm() function
- car.models() now also shows the estimated R2 for each model.


Version 1.0:

- Public release on 8 August 2010.

  This package contains functions for estimation of CAR scores, for
  performing variable selection using CAR scores, and for computing
  corresponding regression cofficients.