K function

Coefficient of divergence

Lubischew's coefficient of divergence (SSMD^2)

K(x, y=NULL, data=NULL, mad=FALSE, na.rm=TRUE)
## S3 method for class 'K'
print(x, ...)
## S3 method for class 'K'
summary(object, ..., num=2)

Arguments

• x: Numeric vector, or formula, or object of the class 'K'
• y: Second numeric vector, or nothing
• data: Data with two columns (in case of formula)
• mad: Non-parametric variant of K (not Lubischew's)
• na.rm: Remove NAs?
• object: Object of the class 'K'
• num: Digits to round
• ...: Additional arguments

Details

One of the effect size measures, Lubischew's K, coefficient of divergence (Lubischew, 1959). Interestingly, the recently invented "striclty standardized mean difference" SSMD (see, for example, "https://en.wikipedia.org/wiki/Strictly_standardized_mean_difference") is just a square root of K.

Returns

K() returns value of K, or nothing. summary.K() returns also magnitude and P, "probability of misclassification".

References

Lubischew A. A. 1959. How to apply biometry to systematics. Leningrad University Herald. N 9. P. 128--136. [In Russian, English abstract].

Author(s)

Alexey Shipunov

Examples

K(1:3, 2:100)
sapply(eq[, -1], function(.x) K(.x ~ eq[, 1]))
summary(K(x17 ~ Species, data=haltica), num=5)