I have applied var model on multiple features ,but var model accepts the stationary data so i have converted non-stationary data to stationary data by doing diff() ,given the data to var model. what my question is ,is their option get back data normal data from stationary to non-stationary in time series?


1 Answer 1


You simply have to add the resulting time series. The cumulative sums are comparable to your original input.

  • $\begingroup$ Thanks for the comments,i am doing difference na , suppose 30 and 20 are my points difference 10 , we have many combinations to get 10 differences how can i remembered original one in this , thats is why my mail probelm , $\endgroup$ Jun 18, 2020 at 6:28
  • $\begingroup$ just give a simple example so that i can understand easily from that get to apply to my code. $\endgroup$ Jun 18, 2020 at 6:41
  • $\begingroup$ Let's say the last data point you have is 30. You predict a differenced series 5, -7, -1. You are predicting their cumulative sum then: 35, 28, 27 $\endgroup$
    – Sean Owen
    Jun 18, 2020 at 15:22
  • $\begingroup$ ok i got it pandas having cumsum() function by using that we can achieve that one ,thanks $\endgroup$ Jun 19, 2020 at 10:23

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.