This file documents software changes since the previous edition.



# FWDselect 2.0 (2015-05-15)

  * the testing procedure have been optimized.

  * test function has been reprogrammed to speed up computation by using
  the parallel package.

  * selection function has been reprogrammed to speed up computation by using the parallel package.

  * a new argument for the number of folds in the cross-validation procedure has been included in the selection function.

  * the diabetes dataset has been added.



# FWDselect 2.0.1 (2015-10-08)

  * the help documentation has been improved for the test function.

  * a new example has been added to the test function in order to explain the speedup = FALSE argument.

  * the error realated with NA's in the selection and qselection function has been solved.



# FWDselect 2.1.0 (2015-12-18)

  * selection function has been reprogrammed to add three more information criteria: Akaike information criterion (aic), AIC with a correction for finite sample sizes (aicc) and bic (Bayesian information criterion).

  * qselection function adds now an asterisk to the returned table. It is shown next to the size of subset which minimizes the information criterion.

  * test function returns now an additional argument ($nvar) which indicates the number of variables that have to be included in teh final model.