1
$\begingroup$

My data contains binary outcome variable and 4 regressors. 2 regressors are binary and 2 are continuous (truncated to whole numbers).

  • Under what circumstances should the continuous variables be standardised in logistic regression?

According to this article variables should be standardised if they are not in the same scale. However, this is slightly confusing for me, we have not talked about standardisation in the context of logistic regression in class.

$\endgroup$
3
  • 3
    $\begingroup$ I can't see a reason to ever standardize them. $\endgroup$ – Frank Harrell Apr 22 '18 at 21:23
  • 3
    $\begingroup$ There is no requirement to standardized predictors for logistic regression, and it is rare that I see this done. $\endgroup$ – Bryan Apr 22 '18 at 21:26
  • $\begingroup$ You don't need to do that. $\endgroup$ – SmallChess Apr 23 '18 at 0:53
2
$\begingroup$

You don't need to standardize for normal logistic regression as long as one keeps units in mind when interpreting the coefficients.

Standardizing can help with interpreting feature importance because then the coefficients should be apples to apples. (ie if your two standardized continuous variables have coeff of 0.01 and 0.7 then you know that the 2nd one is much more important.)

However for regularized logistic regression continuous variables should be standardized for best results.

Is standardisation before Lasso really necessary?

$\endgroup$
1
  • 1
    $\begingroup$ You can also estimate importance without standardizing using likelihood ratio chi-square statistics. $\endgroup$ – Frank Harrell Apr 23 '18 at 12:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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