[R] Is there a book that summarizes different correlation methods?

Peng Yu pengyu.ut at gmail.com
Thu Oct 8 23:22:50 CEST 2009


Hi,

I know what the following correlation methods ("pearson", "kendall",
"spearman") are through wiki. But I'm wondering if there is a good
book that discuss why these metrics are defined and when to use one
coefficient versus the others.

cor.test(x, y,
         alternative = c("two.sided", "less", "greater"),
         method = c("pearson", "kendall", "spearman"),
         exact = NULL, conf.level = 0.95, ...)

cov(x, y = NULL, use = "everything",
    method = c("pearson", "kendall", "spearman"))

cor(x, y = NULL, use = "everything",
     method = c("pearson", "kendall", "spearman"))

I know that "cor(x,y)=cov(x,y)/sqrt(var(x)*var(y))". What is the
relation between cor(), cov(), cor.test() for "kendall" and
"spearman"?

Regards,
Peng



More information about the R-help mailing list