Sorry this is too cryptic for me. Look at [1]; [2] sec 1.2. For details, you may also refer to [4] and to [3] for some history. For non-parallel prototype in R, see R code attached on [1]. You may need also to understand simple SSVD flow in order to understand details of [4]; for which you may want to study N. Halko's dissertation referenced at [1], which is a long thing, but you need pages related to SSVD approach in Mahout.
[1] https://cwiki.apache.org/confluence/display/MAHOUT/Stochastic+Singular+Value+Decomposition [2] https://cwiki.apache.org/confluence/download/attachments/27832158/SSVD-CLI.pdf?version=17&modificationDate=1349984685000 [3] MAHOUT-817 [4] https://issues.apache.org/jira/secure/attachment/12515961/SSVD-PCA%20options.pdf On Mon, Sep 23, 2013 at 6:30 PM, <[email protected]> wrote: > Mr. Dmitriy Lyubimov,. Again I plan to process the results of left > eigenvectors U = AVS ^ (-1) to the PCA algorithm, in order to better value > its variants > > > integrate (SVD) A= USV^(t) to (PCA) X= U^T > > > According to mr. Dmitriy Lyubimov, .. whether in existing mahout who made > ...?...beg an explanation... > > I respectfully
