taucor: Kendall's tau for two variables

In YafeiXu/CopulaModel: CopulaModel: Dependence Modeling with Copulas

Description

Data version of Kendall's tau for two variables, using Knight's algorithm

Usage

taucor(x,y=0,pflag=0)
taub(otab) # Kendall's taub, adjusted for ties for 2-way table

Arguments

x	first variable, ties allowed; n x d data matrix if y=0
y	second variable, ties allowed
pflag	print flag, 0 = silent, 1 = print number of ties, exchange count and other summeries 2 = print sorted data 3 = print sorting iterations
otab	ordinal 2-way table/matrix

Value

tau	Kendall's tau with value in (-1,1); denominator adjusts for ties in case of ordinal variables

References

Knight WR (1966). A computer method for calculating Kendall's tau with ungrouped data. Journal of the American Statistical Association, 61, 436-439.

Examples

set.seed(1234)
x=rnorm(50)
y=.6*x+.8*rnorm(50)
tau=taucor(x,y)
tau2=cor(x,y,method="kendall")  # this is too slow when length(x) is large
cat(tau,tau2,"\n")
z=.5*x+.5*y+sqrt(.5)*rnorm(50)
tau=taucor(cbind(x,y,z))
tau2=cor(cbind(x,y,z),method="kendall")
print(max(abs(tau-tau2)))
# check for two ordinal variables
n=25
set.seed(123)
x1=floor(runif(n,0,3))
x2=floor(runif(n,0,3))
otab=table(x1,x2)
taub(otab) # should be same as next two lines
taucor(x1,x2)
cor(x1,x2,method="kendall")

YafeiXu/CopulaModel documentation built on May 9, 2019, 11:07 p.m.