I am doing the sales forecast. I found the trend and seasonality manually for my time series data. Regressed time series data against the trend and seasonality and found the residuals. The residuals doesnt have autocorrelation(is this expected?). Verified the ACF and PACF plots and found no significant variables. What model to fit for the such residual data? It is normally distributed(confirmed using shapiro test) Any kind of guidance to move forward will be really helpful. I am new to the time series modelling

Please advice. Thanks!


1 Answer 1


Post your data. It is ok to have a mean model and flat forecast. If there are no patterns in the ACF/PACF and no outliers and no changes in seasonality/level/trend/parameters/variance you are ok!

See Rob Hyndman's words on flat forecasts http://robjhyndman.com/hyndsight/flat-forecasts/

  • $\begingroup$ Apologies..I cant share data.. it is restricted..both my ACF and PACF have their first lag variable insignificant.. but 2nd, 3rd is significant..can i do time series modelling with such data? if yes, how many lag variables to consider? $\endgroup$ May 31, 2016 at 1:09
  • $\begingroup$ It sounds like you might need an AR2, but what about outliers...are you looking for them?? You can scale your data by multiplying it by .23080744 for example and you won't be violating anything. $\endgroup$
    – Tom Reilly
    May 31, 2016 at 12:00
  • $\begingroup$ there are no outliers..i ve verified it.. why would you recommend only AR model? in my ACF plot, 2nd and 3rd lag variables significant variables as well.. $\endgroup$ Jun 2, 2016 at 2:34

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.