How to test data is nonstationary or not

[#1]

Hello expert,
I have a 2 column process data table (column 1: date, column 2: value).
How can i test in JMP17 to check whether the data is stationary or not?

Thank you

[#2 · Accepted solution]

By 'stationary' do you mean a pure random walk about some constant mean? There are multitude of time series methods to assess the stationary component of a time series. Some are purely analytical others more 'art'. In the 'art' category I'd start with a simple run chart. What does your eye and process knowledge tell you? Stare at the plot for no more than three seconds. If you don't see a trend or cyclical pattern...chances are very good it's not there and it would probably be safe to presume stationarity. Veering to a bit more analytical approach (but still use your eyes and process knowledge) is to turn on all the Nelson rules and create an appropriate control chart(s). If you violate any of the rules...a good indicator of non stationarity...even rule 1. Lastly, I'd take a look at some of the capabilities in the Time Series platforms in JMP.

[#3]

Thank you @P_Bartell .

From the "art" category, I suspect it is non-stationary so I would like to confirm through some statistical test like ADF test (looking for a p-value).

I've checked the Time Series platform in JMP, but it didn't show p-value for ADF test.

Not sure what does Zero Mean ADF/Single Mean/Trend ADF.

I guess, I have to do it in R/R-Studio.

Thanks anyway !