time series regime detection in R?
Michael While strucchange and segmented look at regime switches triggered by time, tsDyn package investigates regime change triggered by covariate. These are the smooth and threshold AR models, where the regimes will change depending on the past values of the variables, or levels of any other variables. The threshold models are also available in multivariate case, and in the cointegration case. One should also mention in the same vein Markov Models, where regime changes are based on an unobserved Markov process. There is unfortunately no package on CRAN to offer this, but some packages in development on R forge. Matthieu Le 02/02/2012 16:25, Achim Zeileis a ?crit :
On Thu, 2 Feb 2012, Michael wrote:
Hi all, good morning and good evening! Could you please point me to some commands/functions/packages in R commutity that can do regime detection for time series?
The "strucchange" and "segmented" packages provide functionality in this direction. The former considers abrupt shifts (including jumps) in parameters whereas the latter has the additional restriction of looking at continuous segmented functions. If you are in a setting without covariates, the packages "changepoint" and "bcp" may also be of interest.
For example, giving a times series of SP500, and maybe other covariates and macro data, we can answer the following questions: 1. How many regimes are there? 2. Are we currently in a new regime and since when?
A specific application that may be of interest in this direction is implemented in the "fxregime" package (built on top of "strucchange"). See also the accompanying paper: Achim Zeileis, Ajay Shah, Ila Patnaik (2010). Testing, Monitoring, and Dating Structural Changes in Exchange Rate Regimes. Computational Statistics & Data Analysis, 54(6), 1696-1706. doi:10.1016/j.csda.2009.12.005 hth, Z
Thank you!
[[alternative HTML version deleted]]
_______________________________________________ R-SIG-Finance at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance -- Subscriber-posting only. If you want to post, subscribe first. -- Also note that this is not the r-help list where general R questions should go.
_______________________________________________ R-SIG-Finance at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance -- Subscriber-posting only. If you want to post, subscribe first. -- Also note that this is not the r-help list where general R questions should go.