I'm trying to determine correlation between two structured data sets.
Eg:
D1
name: ABC
age: 3
address: xyz
D2
name: ABC
dob: x/t/r
occupation: xyz
Is there a way to use Pearson correlation on structured data instead of variables?
-1
$\begingroup$
$\endgroup$
-
$\begingroup$ This is more appropriate for Cross Validated. $\endgroup$ – Yuval Filmus Sep 7 '17 at 7:18
-
$\begingroup$ What properties do you want the measure of correlation to have? What do you want to use it for? If I propose a possible measure, how do you plan to evaluate it? What criteria will you use? (Note that cross-posting is not allowed, so please don't post the same question on Cross Validated unless you delete this copy first.) $\endgroup$ – D.W.♦ Sep 7 '17 at 22:45
1
$\begingroup$
$\endgroup$
No. Pearson's correlation is a correlation measure between two variables. You might be looking for canonical correlation analysis.
