bqs_rank: Ranking Clusters Quadratic Scores Estimated Via Boostrap

Description

Ranks the scores of clusters methods estimated via boostrap

Usage

bqs_rank(bqsol, rankby = "lq", boot_na_share = 0.25)

Value

An S3 object of class bqs. Output components are those of

bqs. See Value in bqs.

Arguments

bqsol: an object of class bqs obtained from bqs.
rankby: character string specifying how the scored solution are ranked. Possible values are {"lq", "mean", "1se"}. With ="lq" (default), the solutions are ranked by maximizing the estimated lower limit of the of the 1-alpha bootstrap confidence intervalfor the expected score. With ="mean", the solutions are ranked by maximizing the estimated expected score. With ="1se", the solutions are ranked by maximizing the estimated lower limit of the confidence interval for the expected score whose semi-length is equal to a standard error. The expected score's standard error is approximated using the boostrap distribution.
boot_na_share: a numeric value in (0,1). During the boostrapping a method's score is set to NA if the underlying comptutation runs into errors. Methods resulting in more than B * boot_na_share errors are excluded from the comparison

References

Coraggio, Luca and Pietro Coretto (2023). Selecting the number of clusters, clustering models, and algorithms. A unifying approach based on the quadratic discriminant score. Journal of Multivariate Analysis, Vol. 196(105181), 1-20. doi: tools:::Rd_expr_doi("10.1016/j.jmva.2023.105181")

Examples

Run this code

# load data
data("banknote")
dat <- banknote[-1]

## set up methods
## see also help('mset_user')  and related functions 
KM   <- mset_kmeans(K = 3)
GMIX <- mset_gmix(K=3, erc=c(1,100)) 

# combine tuned methods
mlist <- mbind(KM, GMIX)

# perform bootstrap
# change B and ncores to a much larger value in real problems 
res <- bqs(dat, mlist, B = 3, rankby="lq", ncores=1)
res

   
# now change ranking criterion
res2 <- bqs_rank(res, rankby="mean")
res2