1
$\begingroup$

I know normality assumption is not necessary for all of the ARIMA models. My question is that if we have a non-normal time series, is it better to transform it to normal state by transformations like Box-Cox and then use it in the model? what are the pros and cons of these transformation in ARIMA models?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ How do you know that normality assumption is not necessary for all of the ARIMA models? Could you give a reference? $\endgroup$
    – Richard Hardy
    Mar 24, 2015 at 17:05

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.