1

I have plotted ACF and PACF plots for my data, as I am trying to justify my hyperparameters chosen for my ARIMA model.

The data is energy prices every day, with data taken every 30 minutes (48 points every 24 hours) for approximately a total 1 month in length shown below: enter image description here

What is clear is that the ACF plot tails off/gradually decreases and has significant values up to 5, therefore I should have an MA term of 5, however the PACF has significant values up to 8, some negative, some positive which is where I am getting most confused. enter image description here enter image description here

If anyone could take the time to briefly explain it would be VERY much appreciated.

Richard Hardy
  • 67,272
user18376573
  • 11
  • Please describe your data. Would these perhaps be a daily time series of some human-related activity of length 48 days or so? – whuber Apr 10 '22 at 15:53
  • @whuber I edited the post, hopefully it is clearer now – user18376573 Apr 10 '22 at 16:02
  • 1
    You need to look at seasonal differencing first (lag 48). Even then you should check lags out to several days at least. Even then I don't believe the series will be stationary though (day of week effects, day vs night temperature, etc) – Glen_b Apr 11 '22 at 02:43
  • @Glen_b I'm not exactly sure what you mean by looking at seasonal differencing. I did a ADF test, and also looked at rolling statistics, both of which indicated it was stationary data and wouldn't need differencing. Is this the correct approach to check this? – user18376573 Apr 13 '22 at 00:06
  • There's obvious seasonality at exactly the lag you'd expect to see it (lag 48). Even a correctly used test is largely pointless, but the ADF test looks for a lag 1 unit-root effect, not a seasonal effect, so I don't see how that's relevant. See https://stats.stackexchange.com/questions/225087/seasonal-data-deemed-stationary-by-adf-and-kpss-tests#225252 – Glen_b Apr 13 '22 at 00:33
  • @Glen_b Okay I see, thanks a lot for your help, I will look into seasonal differencing – user18376573 Apr 13 '22 at 01:26

0 Answers0