I am working on VAR- time series approach. When defining the lag, it gives me 4. However, when I apply the number of lags=4 to implement the Augmented Duckey filler test to check the stationarity, the null hypothesis will not be rejected for all variables, meaning that my data is not stationary. I tried doing the differencing, which didn't make sense as I had to reach d7 to achieve stationarity for some variables.
Then I tried reducing the lag number in ADF tests to 2, and then I was able to reject the null hypothesis.
I don't know what to do. Please help.
First I'd suggest looking at the raw data and see what it tells you. Just plot the data in the time series in a run chart and see what it looks like. I am afraid people hop to trying to use automated filters without first looking for themselves at the data and making a few unfiltered plots of the data. Dr. Deming would say - it's so simple, just plot the dots. Then that should give some good second-check of what the automated methods are sayig.
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity