1
$\begingroup$

I'm curious if it's useful to collapse numerical dimensions by simply adding them together. For example, I have a data set that has two tightly correlated values. I'd like to trim down the number of dimensions I have so that my ML algorithm has less noise to work with.

Wouldn't it be simple enough to add these two tightly correlated values together and treat them as one dimension?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ It would be simple, sure. But it could destroy crucial information, depending on how those values might be related to whatever you're trying to predict. $\endgroup$
    – whuber ♦
    Commented Jul 13, 2022 at 2:13

0

Reset to default

Your Answer

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