In a multivariate time series, what extra information does performing multivariate change point detection provide than performing univariate change point detection of individual time series? Both of them are likely to give the locations where there is a structural break. What is the advantage of one over the other?
Multivariate system is overall more efficient compared with univariate equation. However, one has to test for if those multivariate conditions are met. You can try both and see the difference.
Dear Laya, econometric applications are not my scientific branch. Please lock for competent colleague of your branch!
In multivariate contexts, there is a joint distribution of the coordinates. When there exists information that is disregarded, one usually loses efficiency, in the sense that earlier detection could be possible (on the average) if the information is used. Furthermore, if coordinates are not independent, acting as if they are would distort the actual operating characteristics.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series