MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

The Kendall tau distance was originally defined as a correlation coefficient. It seems clear to me that every metric function $d$ that is bounded by $b$, induces a correlation coefficient. That is:

Let $d$ be a metric. Then $-d(x,y)/b+1$ is a correlation coefficient.

I wonder if the converse is true, too:

Let $c(x,y)$ be a correlation coefficient. Then $-c(x,y) + 1$ is a pseudometric.

Is the latter statement true or false?

share|cite|improve this question
up vote 3 down vote accepted

The statement is false. Consider the Pearson sample correlation coefficient:

c(x,y) = (x-mean(x)).(y-mean(y))/sqrt(|x-mean(x)|^2*|y-mean(y)|^2)

Here is an example, where the triangle inequality for the distance defined in the question is not satisfied

x = [0.5847 -0.3048 -0.4431 0.5032 -0.3401],

y = [0.2018 0.4547 -0.2230 0.3350 -0.7685],

z = [0.5226 -0.5159 -0.5439 0.3701 0.1671].

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.