Hello,

i would like to have Indpendent Component Analysis in the next version of Igor. I know the "simpleICA" procedure by AG - it works quite well for smaller datasets but it would be nice to have a faster alternative which knows some additional parameters (e.g. choice of contrast function or maybe even a choice for different algorithms like fastICA or Infomax etc). AG informed me that ICA is planned for IP7 but i wanted to state that there is really at least one user who would like this feature :-)
Currently i am trying temporal ICA on highly overdetermined datasets (6400 detectors and approx. 2-50 sources, the length of the recordings is 1000 to 3000 datapoints) and i would love to see the results earlier. If it matters: I need the independent components, the mixing matrix (often called A) and the unmixing matrix (W).

Thank you,

Klaus