gamma.nbest {UCS}R Documentation

Compute Gamma Threshold for N-Best Acceptance Region (gam)


Computes a suitable value of γ such that the acceptance region \{g ≥ γ\} contains exactly n candidates from a given data set.


  gamma.nbest(ds, name, n, jitter=FALSE)


ds a UCS data set object
name name of a generalised association measure (GAM)
n an integer, specifying the number of candidates to be included in the acceptance region
jitter if TRUE, random jitter is added to the coordinates of candidates for computation of the n-best threshold


When jitter=TRUE, the data set ds must contain jitter vectors stored in special variables. Such jitter variables can easily be added with the add.jitter function.


a real number specifying a suitable threshold γ, i.e. the data set ds contains exactly n candidates with a GAM score g ≥ γ (for the specified measure name)

See Also

add.jitter, gam.score, add.gams, gam.iso, builtin.gams


e <- 10^seq(-2, 1, .1)		# 100-best iso-line for UCS data set ds
gamma <- gamma.nbest(ds, "t.score", 100)
o <- gam.iso("t.score", gamma, e=e)

[Package UCS version 0.5 Index]