Testing for stationarity of time series …

As I’m presently stuck in Canberra (waiting on a visa to start work in the UK) I was able to drop in yesterday on a talk at the ANU MSI by Suojin Wang describing his (with a couple of collaborators) new nonparametric test for stationarity of time series in the time domain. (I can’t find the manuscript online so I suppose it’s still under revision? abstract for the talk is here.)  The test statistic—based on the supremum of all sample covariance differences between subsamples of the observed time series selected according to the Walsh functions of degree up to the sample size [if I remember the talk correctly!!]—is provably (theoretically and by numerical example) more powerful than the rival Dette et al test, which is an impressive achievement.  A potential limitation for use in astronomical studies is that at this stage it can’t be used on time series with gaps in coverage, but this is a limitation in common with most such nonparametric tests with proven consistency properties.  Look out for the manuscript (and hopefully the corresponding software, I suppose) …

This entry was posted in Infinite-Dimensional Inference, Non-Parametric. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s