I want to check if my time series variable is stationary.
with no lag order variable is stationary, but with 1 and more lags order process is non-stationary.
what I can conclude: is process stationary or not?
(if it is informative I attach pacf graph)
Dickey-Fuller = -4.6406, Lag order = 0, p-value = 0.01
alternative hypothesis: stationary
Dickey-Fuller = -1.3345, Lag order = 6, p-value = 0.8555
alternative hypothesis: stationary
thank in advance
The PACF plot looks statistically significant. You have conducted the Dickey Fuller Test. How about the KPSS test? This will give you added information.
: A stationary time series is one whose statistical properties such as mean, variance, autocorrelation, etc. are all constant over time.
Ali Hameed Yousif is quite correct. There are various kinds of stationarity. If you are not familiar with this get a good book. I recommend this one:
https://www.amazon.com/Introduction-Analysis-Forecasting-Probability-Statistics/dp/1118745116/ref=sr_1_2?keywords=Montgomery+Time+series+analysis&qid=1565835502&s=books&sr=1-2
There are others but get one with computer package examples. I am an accredited professional statistician (ASA accredited) with a very nice PHD. I found time series to be the most difficult subject to study. If your university has a stat consulting center be sure to use it. Ask lots of questions and you will be ok.
Best, David Booth BTW I wrote my dissertation on Robust time series analysis. You can too.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series