0
$\begingroup$

I have a time series with the following ACF and Partial ACF plots; however, I am a little confused on how to interpret these.

I called the following commends:

acf(x_values)
pacf(x_values)

I believe I can use the $x$-values of my data instead of the residuals. Is that correct?

My thoughts:

  1. The ACF plot shows that the model does a satisfactory job based on the autocorrelation. It is clear that lags $l = 0, 1, 2$ are significant with a couple other lags in the distance (is this a result of cyclical behavior???).
  2. The PACF plot shows more of a weekly cyclical pattern, as each observation refers to one day.

Is what I am saying above correct? Am I producing and interpreting these plots correctly?

enter image description here

Improve this question
$\endgroup$
3
  • $\begingroup$ Have you read through the many existing threads on this that already exist on the site? Did none of them shed any light on this? $\endgroup$
    – gung - Reinstate Monica
    Commented Feb 18, 2021 at 13:38
  • $\begingroup$ @gung-ReinstateMonica yes and my plots are slightly different than the others $\endgroup$
    – 324
    Commented Feb 18, 2021 at 15:02
  • 1
    $\begingroup$ (1) I would not try to interpret barely significant autocorrelations at lags greater than one or two in the first plot. (2) I'm not sure I'm seeing a "weekly" pattern in the second plot, but autocorrelation seems to disappear for lags beyond ten. // You should read related pages as @gung suggests. $\endgroup$
    – BruceET
    Commented Feb 19, 2021 at 0:10

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.