next up previous contents
Next: nanpca.m Up: Some Changes in the Previous: Some Changes in the


Missing values are marked to the data matrix as not-a-number (nan) values.
% Use PCA to get initial values for the sources
[pcasources, pcabias, pcaV, pcaD] = nanpca(data, searchsources);

% values missing from data are collected
missing = sparse(isnan(data));
data(find(missing)) = 0;

Tapani Raiko