1
$\begingroup$

I am using regression model to predict values I have. How can evaluate the accuracy of prediction model (I'd like to see how the accuracy of predicted values).

I found different metrics, but it's only used with discrete data.

  • Classification report
  • Accuracy score
  • Confusion matrix

Now, I can generate the summary using .summary() function. As an example here , it's nearly the same in R/Python;

I'd like to evaluate the performance, how good and accurate the predicted values that I got using GLM (with Binomial families ).

$\endgroup$
1
$\begingroup$

You could use pseudo R-squared measures, such as Nagelkerke, for an overview see:

http://www.ats.ucla.edu/stat/mult_pkg/faq/general/Psuedo_RSquareds.htm

$\endgroup$
  • $\begingroup$ Correct me if I am wrong. If I understand it very well. R-squared is always between 0 and 100%: - 0% indicates that the model is extremely bad. - 100% indicates that the model explains all the variability of the response data around its mean ==> extremely good. $\endgroup$ – user3378649 Apr 25 '14 at 13:16
  • 1
    $\begingroup$ That seems right, except that a % interpretation is more difficult in the pseudo-R-squared case. It's okay in the linear case, where the ratio is explained against total variance. $\endgroup$ – tomka Apr 28 '14 at 12:28

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.