![ACF of my dataset[![][1]](https://i.sstatic.net/Bb8yh.jpg)
Click for ACF image 2
These are my ACF and PACF of my data set . Can any one help me determine whether the data is stationary ? if so What order of AR , MA ?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
2
The periodicity in your (P)ACF plots strongly suggests seasonality of length slightly less than 0.02 of the cycle you defined. Since $7/365\approx 0.019$, I suspect that you have daily data, which you specified as frequency=365. We can't say much more.
Note that it is not very good practice to examine (P)ACF plots any more (the box-jenkins approach). Nowadays, it's better to choose models based on AIC, after determining any necessary differencing using appropriate methods. I suggest you look at ?forecast::auto.arima and FPP2.
If you do have daily data with possible intra-yearly seasonality, you may want to look at our multiple-seasonalities tag.
answered Jul 25, 2018 at 16:27
-
$\begingroup$ Yes its a daily data I used the auto.arima function and it resulted in following Series: myts ARIMA(1,0,0) with non-zero mean Coefficients: ar1 mean 0.2103 236.0485 s.e. 0.0284 65.4923 sigma^2 estimated as 3181549: log likelihood=-10569.71 AIC=21145.41 AICc=21145.43 BIC=21160.65 $\endgroup$– kartikCommented Jul 25, 2018 at 17:13
-
$\begingroup$ Try clarifying that the more relevant seasonal cycle is of length 7, rather than 365:
ts(..., frequency=7). I suspect thatauto.arima()will then give you a seasonal model based on this frequency. $\endgroup$ Commented Jul 25, 2018 at 19:51