Online Supplement to “A Hessian Regularized Nonlinear Time Series Model”

Jie Chen & Xiaoming Huo
September 2008

Click here for an archive file including all the codes for the HRM model. After unpacking, there should be two subdirectories: “Matlab-for-HRM” including Matlab-based scripts that implement HRM; “R-codes” containing R codes for other methods in our comparison.

Below is a list of major functions in the Matlab part: (You can view the file by clicking the filenames)

  • mainfunc_converge.m simulations regarding the convergence theorem.
  • mainfunc_GCV.m simulations regarding generalized cross validation (GCV) principle.
  • mainfunc_pred1.m one-step and iterative two-step predictions with HRM for synthetic examples.
  • mainfunc_pred2.m direct two-step prediction with HRM for synthetic examples.
  • ex_sunspot.m prediction with HRM for the sunspot example.
  • ex_blowfly.m prediction with HRM for the blowfly example.
  • sunspot.mat and blowfly.dat provide the data for sunspot example and blowfly example.

To view usage of all subroutines, you can use the “help FUNCTION-NAME” facility in Matlab. Alternatively, view the HRM Software PDF.

More information regarding functions for other methods (mainly implemented in R) can be seen in HRM Software PDF as well.

We may improve this package from time to time. Newer version will be released here.