Optimal lag for AR in big data
Posted: Fri Jul 28, 2017 2:36 pm
Hi, there. For deciding a benchmark model, I should decide an optimal lag for AR. Raw data is about hourly prices and data size is about more than 35000. The version of Eviews is 7.2
However, Hannan Quinn criteria values decrease as P, lags increase until 23 like the attached image file. In the picture, the second column is HQ and the third is the difference between each. (23 is a max lag Eviews allows in an equation. dlog(price) ar(1).....ar(23) In this case, how can I choose an optimal lag.
Plus, I saw I can use AR(1 to 5) intead AR(1) AR(2) AR(3) AR(3) AR(4) AR(5) if I want to use AR(5). However, it does not work. The function is right? Please help me:)
However, Hannan Quinn criteria values decrease as P, lags increase until 23 like the attached image file. In the picture, the second column is HQ and the third is the difference between each. (23 is a max lag Eviews allows in an equation. dlog(price) ar(1).....ar(23) In this case, how can I choose an optimal lag.
Plus, I saw I can use AR(1 to 5) intead AR(1) AR(2) AR(3) AR(3) AR(4) AR(5) if I want to use AR(5). However, it does not work. The function is right? Please help me:)