I conduct ADF test in level using Eviews, and the data isn't stationer, so I conduct ADF in first difference and the data already stationer. I want to determine the best model for AR and MA to my data, and I conduct the correlogram at first difference, and the result show all the order/bar not exceed the band. What does it mean?
Following an ADF test confirming stationarity after first-order differencing, a correlogram was generated. The observation that all autocorrelation function and partial autocorrelation bars remain within the significance bands indicates the absence of significant autocorrelation.
This outcome suggests that neither a MA nor na AR component is necessary for the model. Consequently, the series is appropriately modelled as a random walk, with the optimal ARIMA model being ARIMA(0, 1, 0).
The final step involves fitting this model and subsequently examining the residuals for white noise properties to validate the model's appropriateness.
Sónia Maria Aniceto Morgado Thank you for your reply, it is really helpful. But, what does a random walk mean? it means I can determine whatever AR and MA order, as long as the model has completed the significant and residuals for white noise, isn't it?
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators