feat.rank.re objects and list of
feat.rank.re objects to form
ftrank.agg object. The main utilities of this function is to concatenate in a single list various results derived from several
feat.rank.re calls in order to facilitate post analysis additional treatments and sorting of the results.
ftdef filed in the result list is a table with 9 columns which are automatically generated to summarise the content of each individual
feat.rank.re. it is also aimed at avoiding confusions if the same method is applied on the same discrimination problem but with different settings, different resampling partitioning or even different data sets. Each column is described as follows:
Unique identifier for each resampling based feature rankings.
Name of the classification technique as specified in the call of
Arguments passed to the FR technique during the call of
Summary of the resampling strategy adopted during the call of
Discrimination task involved. By default, this is equal to the actual levels of the class vector passed to
feat.rank.re separated by
Unique algorithm identifier based on the columns Alg, Arg and Pars so that no confusion is possible with Alg if several rankings have been built with the same FR technique but with different parameters and/or resampling strategy. This column can be modified by the user.
Unique algorithm identifier based on the columns Dis in order to simplified the name of the discrimination task if there are many classes involved and/or class level have a long name. This column can be modified by the user.
Empty column that can be amended to store extra information.
Summary of each
David Enot [email protected]
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
data(abr1) y <- factor(abr1$fact$class) x <- preproc(abr1$pos , y=y, method=c("log10","TICnorm"),add=1)[,110:500] dat <- dat.sel1(x, y, pwise=list(c("1","2"),c("3","2")),mclass=NULL, pars=valipars(sampling="boot",niter=2,nreps=5)) resauc = lapply(dat, function(x) feat.rank.re(x,method="fs.auc")) resrf = lapply(dat, function(x) feat.rank.re(x,method="fs.rf",ntree=100)) mfr=ftrank.agg(resauc,resrf) ### Print out characteristics of each individual FR objects print(mfr$frdef) ### Number of objects in mfr length(mfr$ftrank) ### This is FR object num.1 mfr$ftrank[]
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.