So I've been trying to solve a problem of quantitatively measuring the similarity/difference between groups in my dataset. I am not trying to cluster data to create groups, because the groups are political divisions. I am trying to quantify how similar respective groups are. Here's my thought process:

For each group, calculate a centroid using the n dimensions.

Calculate the Euclidean distances between each point.

So in this 3 dimensional example, if the centroids are at:

Group Coordinates of Centroid
A (1,1,1)
B (2,1,1)
C (10,10,10)

Then the distances would be:

x a b c
a 0 1 15.59
b 1 0 15.03
c 15.59 15.03 15.59

Any critiques or feedback would be very appreciated.

  • $\begingroup$ Find centroid using median. It is robust to outliers. $\endgroup$
    – amol goel
    Nov 16, 2022 at 4:00
  • $\begingroup$ Thanks for the tip @amolgoel $\endgroup$
    – pubb
    Nov 17, 2022 at 5:15


Your Answer

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