correlogram and Dickey fuller test

#1
Hi eveyrone !!

I'm kind of lost in time series analysis. I've never done it before, but I need to make some descriptive analysis of a series. I've run a correlogram and a unit root test, but I don't know what the results mean? How can I determine the stationarity/trend, the seasonal component from those test? my first guess would be that the series is only white noise, but i'd like to confirm it with hypothesis testing :D

Thanks already :)
 
#2
Ok, I found my answer about the unit root test :p

As the ADF test-stat is significantly lower than the critical value, then we reject the hypothesis and the series has no unit root. It it is therefor stationary at 1% 5% and 10% significant level.
 

noetsi

Fortran must die
#3
Essentially the null of this test (you are actually looking at the augmented text) is that the data is non-stationary. So if you reject the null (i.e., have values below your alpha level) your data is stationary. Note that stationarity can involve both seasonal and non-seasonal elements or both.

Good luck at analyzing correlograms. I am working with those and it is never easy. Duke University has an excellent website that makes a series of suggestions on how to interpret ACF and PACF. Needless to say different authors disagree on the rules.