Neuralnets for Multivariate And Time Series Analysis (NeuMATSA)

MATLAB codes for:

  • Nonlinear Principal Component Analysis (NLPCA)
  • Nonlinear Canonical Correlation Analysis (NLCCA) (**** Bug report (2008/1/27) **** ).
  • Nonlinear Singular Spectrum Analysis (NLSSA)

    The latest release (version 5.0) became available in Oct. 2007. [The major improvements over the previous version are: (1) The appropriate weight penalty parameters are now objectively determined by the codes. (2) Robust options have been introduced in the codes to handle noisy datasets containing outliers.]

    The programs are free software, under the terms of the GNU General Public License as published by the Free Software Foundation. The codes are written in MATLAB and use its Optimization Toolbox.

    First download the manual:

  • Hsieh, W.W., 2008. Neuralnets for Multivariate And Time Series Analysis (NeuMATSA): A User Manual (in PDF format).

    Next download the 2004 general review paper and more recent paper(s) of relevance:

  • Hsieh, W.W., 2004. Nonlinear multivariate and time series analysis by neural network methods. Reviews of Geophysics, 42, RG1003, doi:10.1029/2002RG000112. (reprint with typos corrected in PDF)

  • Hsieh, W.W., 2007. Nonlinear principal component analysis of noisy data. Neural Networks, 20: 434-443. DOI 10.1016/j.neunet.2007.04.018. (preprint in PDF)

  • Cannon, A.J. and W.W. Hsieh, 2008. Robust nonlinear canonical correlation analysis: application to seasonal climate forecasting. Nonlinear Processes in Geophysics, 15: 221-232. (preprint in PDF)

    Some of the papers written by our group referenced in this review paper can be downloaded at the site http://www.ocgy.ubc.ca/~william/pubs.html.


  • For Nonlinear Complex Principal Component Analysis (NLCPCA), Sanjay Rattan (e-mail: "srattan" followed by "@ualberta.ca") converted the NLPCA code to complex variables. There is no written manual other than a file manual.m attached to the Matlab codes. The relevant publication is:
    Rattan, S.S.P. and Hsieh, W.W., 2005. Complex-valued neural networks for nonlinear complex principal component analysis. Neural Networks, 18: 61-69, DOI:10.1016/j.neunet.2004.08.002. (preprint in PDF).

    Finally download the codes.