I have two arrays of 2D points and I need to estimate their correlation. What formula should I use?

Example of arrays:

$$X: ((1,5),(2,5),(1,7),(4,1)),$$

$$Y: ((3,4),(1,6),(4,6),(4,3)).$$

  • 1
    $\begingroup$ A reference is Tobler, W.R. 1965. Computation of the correspondence of geographical patterns. Papers in Regional Science 15: 131–139. doi: 10.1111/j.1435-5597.1965.tb01318.x $\endgroup$
    – Nick Cox
    Oct 26, 2013 at 0:12
  • 2
    $\begingroup$ Another reference (much more recent and accessible) jstatsoft.org/v52/c01/paper $\endgroup$
    – Nick Cox
    Oct 26, 2013 at 0:36
  • 1
    $\begingroup$ Brownian correlation? It can be computed between arrays of any, even different dimensionality. $\endgroup$
    – ttnphns
    Oct 26, 2013 at 6:12
  • $\begingroup$ Thank you for your replies! As for now I started trying Brownian correlation and RV coefficient. The coefficient from Tobler (if I understood correctly) is not symmetrical and it seems like a disadvantage for me, so I didn't try it. $\endgroup$ Oct 26, 2013 at 20:30
  • $\begingroup$ Why not distance covariance or distance correlation? en.wikipedia.org/wiki/Distance_correlation#Distance_covariance $\endgroup$ Aug 24, 2016 at 6:35

1 Answer 1


The Pyramid Match Kernel is a flexible technique for estimating a normalized correspondence strength between multidimensional point sets. It is analogous to $R^2$ in many ways (symmetric, values in $[0,1]$), and could suit your purposes.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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