Description Usage Arguments Value Note Author(s) See Also Examples
Wrapper that aggregates results obtained from a feature ranking resampling strategy.
If p values have been calculated by
the feature ranking technique on the overall dataset, adjusted p-values by one of
the methods available in p.adjust
are also returned.
1 | fs.summary(res1,res2,padjust="fdr",sorting=TRUE)
|
res1 |
A list returned from |
res2 |
A list returned from |
padjust |
One of the methods in |
sorting |
Should the results be sorted according to the feature ordering calculated on the overall data? |
A matrix of statistics. For details, see Note
below.
The output matrix with number of rows corresponding to the number of variables and number of columns to:
Feature ranking quantity computed on the whole dataset
Feature rank in decreasing order of saliency
p-value if available with feature ranking technique (optional)
adjusted p-value if p-value available (optional)
average feature rank across every resampling steps
standard deviation of the feature rank across every resampling steps
pseudo p-value calculated from the resampling strategy
David Enot and Wanchang Lin dle, wll@aber.ac.uk.
feat.rank.re
, fs.mrpval
, p.adjust
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 | ## load abr1
data(abr1)
y <- factor(abr1$fact$class)
x <- preproc(abr1$pos , y=y, method=c("log10","TICnorm"),add=1)[,110:500]
## Select classes 1 and 2
dat <- dat.sel(x, y, choices=c("1","2"))
x <- dat$dat[[1]]
y <- dat$cl[[1]]
## partitioning
pars <- valipars(sampling="boot",niter=2,nreps=5)
tr.idx <- trainind(y,pars=pars)
## multiple rankings using AUC
z <- feat.rank.re(x,y,method="fs.auc",pars = pars,tr.idx=tr.idx)
## Compute stability mr-p value using the 75% worst features as irrelevant
res <- fs.mrpval(z,qtl=0.25)
## print content of res
names(res)
res.1 <- fs.summary(z, res, sorting=TRUE)
## Print the 10 best features
print(res.1[1:10,])
##### Example of output with a feature ranking technique that returns p-value
z <- feat.rank.re(x,y,method="fs.welch",pars = pars,tr.idx=tr.idx)
res <- fs.mrpval(z,qtl=0.25)
names(res)
## p-value correction with fdr
res.1 <- fs.summary(z, res, padjust = "fdr", sorting=TRUE)
## Print the 10 best features
res.1[1:10,]
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.