Distance-Based Statistics

Prediction methods where explanatory information is coded as a matrix of distances between individuals. Distances can either be directly input as a distances matrix, a squared distances matrix, an inner-products matrix or computed from observed predictors.


Version 1.0.5 [2017-10-17]

o Corrected a minor error in the formulas for aic and bic in the dblm code.

Version 1.0.4 [2014-12-12]

o New summary format for ldbglm objects. Two new functions have been defined: summary.ldbglm and print.summary.ldbglm

o The use of "offset" now is allowed as an additional parameter in dbglm

Version 1.0.3 [2013-11-22]

o Corrected maintainer's name format as instructed.

o Function as.D2 now only changes the type of the input object and do not makes its squared version.

o The character "_" has been eliminated of the package. It has been replaced by "." as is the case of:

- full_search -> full.search (dblm)
- user_h -> user.h (ldblm and ldbglm)
- scores_comps -> scores.comps (plot.dbplsr)
- h_opt -> h.opt (ldblm and ldbglm)
- ncomp_opt -> ncomp.opt (dbplsr)
- type_var -> type.var (predict.dblm, predict.ldblm, predict.dbglm, 
  predict.ldbglm, predict.dbplsr)
- type_glm -> type.glm (plot.dbglm)

o The name of the argument "method" in functions ldblm and ldbglm has been changed by "method.h" when it refers to the bandwidth selection.

o The name of the argument "type" in predict.dbglm and predict.ldbglm has been changed by "type.pred". The name the argument "type.glm" in plot.dblm has been changed by "type.pred". In both cases the name has been changed to unify the name of arguments refered to the same concept.

o The argument "method" in function dbglm now includes five options: "Eff.rank", "rel.gvar", "AIC", "BIC" and "GCV" instead of two (eff.rank and rel.gvar). As a consequence in plot.dbglm it is possible to choose "which = 6" as in plot.dblm, to visualize the minimum achieved with the criterion specified in "method".

o The summary command of function dbglm also prints the vale of the BIC and GCV crteria of the final DB-GLM.

o The default value for "method" in function dbglm is now "GCV". The default value for "full.search" in function dbglm is now "TRUE".

o The argument "range.eff.rank" has been added in function dbglm.

o The argument method.h of function ldbglm does not includes now the option "OCV".

o The value of function dbglm now includes too: call, dev.resids, varmu, bic.model and gcv.model.

o It has been applied when it has been necessary an estandarization of variables taking into account the original weights.

o In the programming it has been changed: - the name of "Llindar" by "threshold"; - the value 9999999 by Inf.

o Misprints have been modified in the help file and in the summary, print and plot commands.

Version 1.0.2 [2012-05-18]

o The calls of functions dblm, ldblm, dbglm and ldbglm have been unified by eliminating .formula, .dist, .D2, .Gram of its names. The information related to the different usages is now indicated by the arguments.

o Misprints have been modified in the help file.

Version 1.0.1 [2011-06-30]

o First release on CRAN

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


1.0.5 by Josep Fortiana, 4 years ago

Browse source code at https://github.com/cran/dbstats

Authors: Boj , Eva <[email protected]> , Caballe , Adria <[email protected]> , Delicado , Pedro <[email protected]> and Fortiana , Josep <[email protected]>.

Documentation:   PDF Manual  

Task views: Functional Data Analysis

GPL-2 license

Depends on cluster, pls

Suggests proxy

See at CRAN