[mlpack-svn] [MLPACK] #298: Enhancement of linear regression library

MLPACK Trac trac at coffeetalk-1.cc.gatech.edu
Wed Sep 11 16:17:40 EDT 2013


#298: Enhancement of linear regression library
--------------------------------+-------------------------------------------
  Reporter:  sumedhghaisas      |        Owner:  rcurtin     
      Type:  enhancement        |       Status:  accepted    
  Priority:  major              |    Milestone:  mlpack 1.0.7
 Component:  mlpack             |   Resolution:              
  Keywords:  linear_regression  |     Blocking:              
Blocked By:                     |  
--------------------------------+-------------------------------------------

Comment (by rcurtin):

 Hello Sumedh,

 I apologize again for the slow response.  I had a paper deadline yesterday
 and was very busy preparing for it...

 Which dataset are you referring to for your check for ridge regression?
 Also, I think another good test for ridge regression is to make sure that
 it solves the problem it was intended to solve: when X^T X isn't
 invertible.  So we should design some data matrix that has a non-
 invertible covariance matrix, and then choose a ridge regression parameter
 large enough so that adding to the diagonal is enough to make it
 invertible and the regression will be successful.  Does that make sense?

-- 
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/298#comment:9>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed at Georgia Tech.


More information about the mlpack-svn mailing list