The method irank compares ranks using the same vector as reference.
irank_against returns integer ranks, that values from x would assume if (individually)
inserted into v. frank_against acts analogously, returning fractional ranks.
irank_against(x, v, omega = 0, increasing = FALSE, na.rm = FALSE)
frank_against(x, v, omega = 0, increasing = FALSE, na.rm = FALSE)numeric query vector.
numeric reference vector.
numeric value in [0,1], defining how ties in x (if any) are handled; default is 0. See Details.
logical; if FALSE (default), then large elements in x receive a small rank. Otherwise, large elements in x receive a large rank.
logical; if TRUE, then NA's are removed from x. Default: FALSE.
Numeric vector of the same length as x containing the integer (for irank_against) or fractional (for frank_against) ranks.
It's useful to think about frank_against(x,v) as a generalization of Empirical Cumulative
Distribution Function, created for v and evaluated for points in x.
frank_agaist(x,v,increasing=TRUE,omega=1) is identical
to ecdf(v)(x).
increasing switches the inequality sign in ECDF definition from
\(F_V(t) = \hat P(V <= t)\) to \(\hat P(V >= t)\).
omega=0 introduces the strict inequality (\(\hat P(V < t)\) instead of \(\hat P(V <= t)\)).
Any omega in between is a weighted average of the cases omega=1 and omega=0.
Finally, irank_against is equal to frank_against multiplied by the length(v).
This particular choice of default parameters was made for compatibility with default parameters of
irank and frank. irank(x) is always equal to irank_against(x,x) and frank(x) is always equal to frank_against(x,x).
irank_against(1:10, c(4,4,4,3,1,10,7,7))
#> [1] 8 8 7 4 4 4 2 2 2 1