Hello,
I have a fairly simple model I am trying to solve using Lasso:
AX + B = 0
Where:
A is an [nxn] matrix
X is an [nxm] matrix
and B is an [nxm] matrix
And I am attempting to solve for A, given X and B.
Using a pseudo-inverse, this is quite easy to solve - but I am not really
sure how to solve this using Lasso or any of the other linear solvers.
Most of the solvers available in sklearn assume that you are trying to
solve a problem of type:
Xw - y
Given X and y.
Is there a way to move the problem around (ideally, a method that is
somewhat numerically robust) so that I can use the existing solvers?
The second issue - once I've obtained the parameters from the regression,
what's the correct way to calculate the covariance matrix to obtain an
estimate of the parameters? Does the choice of linear model I've made to
estimate the parameters influence the proper methodology I should use to
estimate the covariance matrix?
Thank you very much,
Federico
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
