Assume I have items: a, b, and c. And I calculated the similarities between each pair using two different measures, and I got the following similarity score metrics:

using measure 1, the results are:

    a       b       c
a   1       0.5     0.1 
b   0.4     1       0.3
c   0.1     0.2     1

and by using measure 2, the results are:

    a       b       c
a   1       0.9     0.6
b   0.1     1       0.2
c   0.3     0.7     1

Is there an approach to combine both score matrixes into a single matrix, does using average will be a good choice? Have you seen such papers?

To put more context, my problem is finding similarities between cities, so the two measures are totally different. For example, measure 1 tries to find similarities based on population, while measure 2 tries to find similarities based on street layout.

  • $\begingroup$ This is the proverbial comparison of the weights of apples with counts of oranges: we have no information about how to make tradeoffs between the measures and we don't even know whether the similarity metrics have any relationship to each other. Your question needs much more information to be answerable. $\endgroup$
    – whuber
    May 16, 2022 at 22:28

1 Answer 1



Imagine a tall skinny person going to get pants which have a length and width measurement but not finding anything that fits because on average taller people are also wider.

  • $\begingroup$ hhhh funny answer. Anyway, I came across this paper. I will read to see if it is relevant eprints.qut.edu.au/18348 $\endgroup$ May 16, 2022 at 18:20

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