search

myriamlr / exposition-family

Automatically exported from code.google.com/p/exposition-family
0 stars 0 forks source link

epCA - system is singular (via Lapack) #31

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
With the IG data (at the time of writing this), epCA throws this error:

Error in solve.default(diag(D)) : 
  Lapack routine dgesv: system is exactly singular
In addition: Warning message:
In sqrt(eigOut$values) : NaNs produced

For both Hellinger and standard CA. FactoMineR and equivalent code in Matlab 
does not.
This is the first instance of this, and the only time I have seen this. It is 
only replicable with this data set thus far, and regardless of hellinger=.

Original issue reported on code.google.com by DerekBea...@gmail.com on 29 Nov 2012 at 7:08

GoogleCodeExporter commented 9 years ago 
This happened on a correct and incorrect format of the data (see your code).

Original comment by DerekBea...@gmail.com on 29 Nov 2012 at 7:24

GoogleCodeExporter commented 9 years ago 
Solution is to drop the solve() function and go with a direct invert.

Original comment by DerekBea...@gmail.com on 21 Dec 2012 at 12:56

GoogleCodeExporter commented 9 years ago 
Fixed!

Original comment by DerekBea...@gmail.com on 26 Mar 2013 at 1:03