1
$\begingroup$

I have extracted the Pearson correlation Coefficients between the labels and features. Can I multiply the final correlation score for each score by 10 to make it higher to investigate it with machine learning?

$\endgroup$
9
  • 1
    $\begingroup$ (1) What do you hope to achieve by the rescaling? (2) Why are you interested in the correlation between the labels and the features? Why don't you directly model the labels, with the features as predictors? $\endgroup$ Commented Feb 22, 2022 at 13:11
  • $\begingroup$ By labels do you mean a categorical variable? $\endgroup$ Commented Feb 22, 2022 at 13:11
  • 2
    $\begingroup$ I've posted an answer, though I am not at all clear what you want to do with these numbers. It might be best to say what you're trying to do. $\endgroup$
    – Dave
    Commented Feb 22, 2022 at 13:18
  • 1
    $\begingroup$ It would depend on how you're achieving that, but if that's not what you're doing we should probably leave that aside. $\endgroup$
    – Glen_b
    Commented Feb 23, 2022 at 10:29
  • 1
    $\begingroup$ The correlation is whatever it is. You don’t get to pick it any more than you get to pick how tall you are or how fast you can run. $\endgroup$
    – Dave
    Commented Feb 23, 2022 at 10:33

1 Answer 1

1
$\begingroup$

SURE

That's a conversion of units, analogous to expressing in decimeters instead of meters.

However, I have my doubts that this would make a difference. When it might matter is if you need to present to an audience that would rather see percentages (so multiplying by $100\%$, not just $10$) than decimal numbers less than $1$ (e.g., $26\%$ instead of $0.26$).

$\endgroup$

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.