So, at present, I am studying for a Time Series midterm.

After looking around for a while, I had difficulty finding a unified table describing the behaviour of different time-series models based on their behaviour as a function of their ACF, PACF, Periodogram or CCF plots. As such, I was wondering if we might be able to together put together such a table (e.g. below), plausibly with examples? Here's what I have so far:

$$ \begin{align} \text{Model} \ & || \quad \text{ACF} &|\quad &\text{PACF} &|\quad \text{Periodogram}\\ AR(p) \ & || \quad \text{Tails off} &|\quad &\text{Cuts off after lag } p &|\quad \text{?} \\ MA(q)\ & || \quad \text{Cuts off after lag } q &|\quad &\text{Tails off} &|\quad \text{?} \\ ARMA(p,q) \ & || \quad \text{Tails off} &|\quad &\text{Tails off} &|\quad \text{?} \\ others \ & || \quad \text{cyclic at lag } \frac{1}{\omega} &|\quad &\text{?} &|\quad \text{spike at }\omega \\ \vdots \ & | \ \quad \cdots \end{align}$$

  • $\begingroup$ Need to use the self study tag $\endgroup$ – Michael R. Chernick Mar 6 '17 at 5:31

For AR, MA, and ARMA, the peaks in the periodogram will depend on the parameters $\phi$s.

For instance, in the case of AR(2), the spectral shapes are determined by: $|\phi_1(1-\phi_2)|<|4\phi_2|$. When $\phi_1^2+4\phi_2\leq0$, the periodogram presents a peak spectrum, otherwise it will displays a long shape. You can find a quite informative diagram for the special case AR(2) in the following book, page 337:

Cryer, J. D., & Chan, K.-S. (2008). Time Series Analysis: With Applications in R. Springer Science & Business Media.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.