While modeling Lynx series after applying log10 transformation, I get ARIMA(2,0,3) as the best fitted model using auto.arima in R. But I see Zhang 2003 (Time series forecasting using a hybrid ARIMA and neural network model, G. Peter Zhang) specifies, after applying log10 transformation, ARIMA(12,0,0) as the best fitted model using MATLAB. This has also been followed up by many researchers Khandelwal et al. 2015, Adhikari R and Agrawal R.K 2011, etc. Can some one clarify if R and MATLAB can provide different models as the best fitted model(s)?
It depends on your criteria and the variables you use. Some references are given in the attached paper. Best wishes David Booth
Thank you David Eugene Booth for your answer.
Thank you Ette Etuk for your answer.
- Badges
- Science topic
- Similar topics
- Filing