1
$\begingroup$

I have different datasets and I want to find out the features that are similar among the datasets. The datasets are of varying sizes. example: dataset1 has columns a,b,c,d,e dataset2 has columns m,n,o,p,q. We as human some how see that column a similar to m , c is similar to n and e is similar to q. But how to achieve this from ML?

$\endgroup$
3
  • 2
    $\begingroup$ Can you expend what you mean by similar ? similar distribution ? similar behavior relatively to a target ,(do you have a target ?) $\endgroup$ Jun 26, 2022 at 14:52
  • $\begingroup$ Similar Distribution $\endgroup$ Jun 27, 2022 at 6:01
  • $\begingroup$ You could apply some distribution distance measure between pairs of features, like KL. $\endgroup$
    – Erwan
    Jun 30, 2022 at 8:13

2 Answers 2

1
$\begingroup$

To find clusters of features, you'll have to define a distance metric. A distance metric is a numeric value of how close together the values of the data are. Distance metrics depend on how the data is measured. Common ways to measure data are binary, nominal, and interval/ratio. Distance metrics can include distributions, Kullback–Libeler (KL) divergence and earth mover's distance.

$\endgroup$
0
$\begingroup$

It depends on what you are doing. I would consider:

  • K-means clustering
  • Analysis of covariance
  • PCA loadings
$\endgroup$

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.