[mlpack-svn] [MLPACK] #292: PCA failed with scale option.

[MLPACK Trac]

#292: PCA failed with scale option.
----------------------+-----------------------------------------------------
  Reporter:  marcus   |        Owner:        
      Type:  defect   |       Status:  closed
  Priority:  trivial  |    Milestone:        
 Component:  mlpack   |   Resolution:  fixed 
  Keywords:           |     Blocking:        
Blocked By:           |  
----------------------+-----------------------------------------------------
Changes (by rcurtin):

  * status:  new => closed
  * resolution:  => fixed


Comment:

 ccov() is used to calculate the covariance matrix on column-major data and
 that calculation is the same regardless of the centering of the data.

 r15270 fixes the bug in that line introduced in r12090.  That changeset
 doesn't actually make any sense, and after thinking about it for a while I
 found how to scale to unit variance after calculating the covariance
 matrix.  I commented what I did thoroughly.

 r15269 introduces a simple test that (I think) ensures that the scaling
 parameter is working correctly.

 I took a look to see if these changes are applicable to Kernel PCA but
 there is no scaling parameter there so I did not need to make any changes.

-- 
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/292#comment:1>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed at Georgia Tech.