0

I used fcm function in matlab with my dataset which is trainset. now I have some testset.

[center,U,obj_fcn] = fcm(train ,cluster_number);

after clustering trainset , how can I ask the degree of membership and most probable cluster for my testset? does it have function ?

patra
  • 9
  • 4
  • @Anony-Mousse Probably the distance between new data and centers which has been determinated . do U have another idea? – patra Mar 31 '16 at 07:07
  • See the FCM paper for how points are assigned to clusters. It's fuzzy, a point can belong to more than one cluster. – Has QUIT--Anony-Mousse Mar 31 '16 at 07:11
  • @Anony-Mousse for example if membership degree for a cluster is 0.68 it is the most probable and it mostly belongs to this cluster. but in fuzzy clustering , a data belongs to all cluster but weak and strong( depend on membership degree) – patra Mar 31 '16 at 07:13
  • @Anony-Mousse yes , you 're right . but how to determine these degrees for a new data test? – patra Mar 31 '16 at 07:14
  • Exactly as you do for the old data set... – Has QUIT--Anony-Mousse Mar 31 '16 at 17:20

0 Answers0