[Scipy-tickets] [SciPy] #447: Give option on what to do when kmeans "forgets" one cluster

SciPy scipy-tickets@scipy....
Tue Jul 3 06:30:12 CDT 2007


#447: Give option on what to do when kmeans "forgets" one cluster
---------------------------+------------------------------------------------
 Reporter:  cdavid         |        Owner:  cdavid                                
     Type:  enhancement    |       Status:  closed                                
 Priority:  high           |    Milestone:  pymachine: cleaning existing toolboxes
Component:  scipy.cluster  |      Version:                                        
 Severity:  normal         |   Resolution:  fixed                                 
 Keywords:                 |  
---------------------------+------------------------------------------------
Changes (by cdavid):

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

Old description:

> in the kmean algorithm, it may happen that one cluster do not have any
> samples anymore, which means the cluster is "lost". For now, a warning is
> sent when this happen; this is fine when using kmeans2 directly, but can
> be a problem when used by other algorithms who may need to catch the
> error. Maybe as a parameter  ?

New description:

 in the kmean algorithm, it may happen that one cluster do not have any
 samples anymore, which means the cluster is "lost". For now, a warning is
 sent when this happen; this is fine when using kmeans2 directly, but can
 be a problem when used by other algorithms who may need to catch the
 error. Maybe as a parameter  ?

 fixed in r3143

-- 
Ticket URL: <http://projects.scipy.org/scipy/scipy/ticket/447#comment:3>
SciPy <http://www.scipy.org/>
SciPy is open-source software for mathematics, science, and engineering.


More information about the Scipy-tickets mailing list