Prev: Advanced PCA Problem: Application to Timeseries
Next: making mex-file compilation dependant on the compiler
From: Avishay on 17 Feb 2010 09:45 Hi, I'm clustering objects and want to be able how "good" (homogeneous) is the clustering. inconsistent() (described in http://www.mathworks.com/access/helpdesk/help/toolbox/stats/inconsistent.html) provides me with an "Inconsistency coefficient" which seems to be what I'm looking for. However there seems to be a mistake in the documentation. Shouldn't Y(k,4) = (z(k,3) - Y(k,1))/Y(k,2) be Y(k,4) = (Y(k,3) - Y(k,1))/Y(k,2) ? I don't understand where the "z" came from, and since the table above describes the 4 columns of Y I assume it should be "Y" instead of "z". If indeed it is meant to be "z", can you explain me what this value describes?
From: Oleg Komarov on 17 Feb 2010 09:58
"Avishay " <al(a)cs.technion.ac.il> wrote in message <hlgve3$qdn$1(a)fred.mathworks.com>... > Hi, > > I'm clustering objects and want to be able how "good" (homogeneous) is the clustering. > inconsistent() (described in http://www.mathworks.com/access/helpdesk/help/toolbox/stats/inconsistent.html) > provides me with an "Inconsistency coefficient" which seems to be what I'm looking for. > However there seems to be a mistake in the documentation. Shouldn't > Y(k,4) = (z(k,3) - Y(k,1))/Y(k,2) > be > Y(k,4) = (Y(k,3) - Y(k,1))/Y(k,2) > ? > I don't understand where the "z" came from, and since the table above describes the 4 columns of Y I assume it should be "Y" instead of "z". > > If indeed it is meant to be "z", can you explain me what this value describes? type on the command line: help inconsistent Oleg |