Home

Package tsDyn: Wiki

Questions? Look at the small FAQ page or on the archive of the Mailing list, where you can also ask directly.

tsDyn is a R package for modelling non-linear time series. It features following models:

Univariate models:

• AR: standard linear AR (auto-regressive)
• SETAR: self-exciting threshold AR
• LSTAR: Logistic smooth transition AR
• NNET: neural-network
• AAR: additive AR

Bi/multivariate models:

• Linear VAR and VECM
• Threshold VAR (TVAR) and VECM (TVECM), with threshold cointegration (bi-variate systems only)

Non-linearity tests:

• Univariate test of linearity against:

  -linearity against SETAR: test of Hansen 1999, setarTest

  -linearity against LSTAR:

• Bivariate test of cointegration against threshold cointegration (Hansen and Seo 2002) TVECM.HStest()

• Bivariate test of no cointegration against threshold cointegration (Seo 2006) TVECM.SeoTest()

Source code

The code is hosted on Matthieu's github, see https://github.com/MatthieuStigler/tsDyn/. And some cases, you will be interested in the development version, which is hosted under the branch Dev94: https://github.com/MatthieuStigler/tsDyn/tree/Dev94. To install from Dev94, do (will need to install Rtools on Windows):

library(devtools)
install_github("MatthieuStigler/tsDyn", ref="Dev94", subdir="tsDyn")