"My own" factor waveform plots (rescaled to microvolts)

Help
2014-07-27
2014-07-29
  • Daniel Silva
    Daniel Silva
    2014-07-27

    Hi!

    I have results of a temporospatial PCA and would like to make my own factor waveform plots outside EP Toolkit, using the microvolt rescaled data from which the plots that I find in the "View" pane are made. The problem is that i can't find those (rescaled) data in the matlab workspace.

    I thought that I could, by myself, multiply scores, loadings, and standard deviations as described in Dien and Frishkoff (2005). However, it would be pretty hard since I have to figure out, for example, which line in the FacScr matrix is associated to which electrode, subject and condition (in order to calculate the standard deviation of the temporal scores at each electrode).
    Is there an easier way?

    The main reason why I need to make my own plots is that I want to compare more than four waveforms on a single plot (I believe there is no way to have more than four conditions represented in the same "Waves" (or "Topos") window).

    Cheers,

    Daniel

     
    • Joe Dien
      Joe Dien
      2014-07-27

      The easier way is to just take the EPdata.data matrix (which for PCA data is the factor scores) and multiply it by the .facVecS or facVecT or facVecF matrix (which contains the factor loadings already multiplied by the standard deviations). I do it this way to save on disk space.

      Cheers!

      Joe

      On Jul 27, 2014, at 3:15 PM, Daniel Silva silvadmr@users.sf.net wrote:

      Hi!

      I have results of a temporospatial PCA and would like to make my own factor waveform plots outside EP Toolkit, using the microvolt rescaled data from which the plots that I find in the "View" pane are made. The problem is that i can't find those (rescaled) data in the matlab workspace.

      I thought that I could, by myself, multiply scores, loadings, and standard deviations as described in Dien and Frishkoff (2005). However, it would be pretty hard since I have to figure out, for example, which line in the FacScr matrix is associated to which electrode, subject and condition (in order to calculate the standard deviation of the temporal scores at each electrode).
      Is there an easier way?

      The main reason why I need to make my own plots is that I want to compare more than four waveforms on a single plot (I believe there is no way to have more than four conditions represented in the same "Waves" (or "Topos") window).

      Cheers,

      Daniel

      "My own" factor waveform plots (rescaled to microvolts)

      Sent from sourceforge.net because you indicated interest in https://sourceforge.net/p/erppcatoolkit/discussion/767176/

      To unsubscribe from further messages, please visit https://sourceforge.net/auth/subscriptions/


      Joseph Dien,
      Senior Research Scientist
      Maryland Neuroimaging Center
      University of Maryland

      E-mail: jdien07@mac.com
      Phone: 202-297-8117
      http://joedien.com

       
      Attachments
  • Daniel Silva
    Daniel Silva
    2014-07-29

    Dear Dr. Dien,

    so if I understand it correctly, in order to plot some spatiotemporal factors, I have to multiply the loadings in FacVecS by the loadings in FacVecT and then by the scores in Epdata.data. After some testing it seems to work quite well.
    Thank you very much.
    Sincerely,

    Daniel

     
    • Joe Dien
      Joe Dien
      2014-07-29

      That’s right! Glad to hear it’s working well for you!

      Joe

      On Jul 29, 2014, at 2:59 PM, Daniel Silva silvadmr@users.sf.net wrote:

      Dear Dr. Dien,

      so if I understand it correctly, in order to plot some spatiotemporal factors, I have to multiply the loadings in FacVecS by the loadings in FacVecT and then by the scores in Epdata.data. After some testing it seems to work quite well.
      Thank you very much.
      Sincerely,

      Daniel

      "My own" factor waveform plots (rescaled to microvolts)

      Sent from sourceforge.net because you indicated interest in https://sourceforge.net/p/erppcatoolkit/discussion/767176/

      To unsubscribe from further messages, please visit https://sourceforge.net/auth/subscriptions/


      Joseph Dien,
      Senior Research Scientist
      Maryland Neuroimaging Center
      University of Maryland

      E-mail: jdien07@mac.com
      Phone: 202-297-8117
      http://joedien.com