Recently I am doing a time series ARIMA modeling using water consumption data and I plotted the time series plot of data so according to this figure Does this time series data have trend or seasonal pattern or both?
Thanks
Hi, I am surely a beginner in time series but the most obvious thing seems to be the structural break 2015. Before the trend seems negative linear, and after this, it seems nonlinear.
(Actually, I am myself currently struggling to find out whether it is possible to model a non-linear trend AND a structural break)...
Best,
Holger
You yould check the autocorrelation and the Furier transform to check for sesonal variation.
For me it looks like a random walk. I don't see evidence for any structural break here. In a random walk, it's not too unlikely to have some adjacent points being close together somewhere, and this may look like a local peak or like a jump. Interpreting this as a structural break seems to be an over-interpretation. If there is something moreknown about the time course, one could simulate data to get an approximate satistical significance of such "breaks".
No obvious periodic patterns are apparent in the run sequence plot. I suggest that you proceed to unit root analyses using both Augmented Dickey-Fuller (ADF) and Philips-Perron (PP) tests (with NO DRIFT) respectively for proper determination of the order of integration of the series. You can download my work on ARIMA titled"ON THE FORECAST OF CEREALS PRODUCTION IN NIGERIA" for more information on ARIMA modeling.
Thanks
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling