I have a data matrix with 1003001 rows and 3 coloums. I did the clustering with Fuzzy C-Means (FCM) and Gustafsson Kessel (FGK) in Matlab software vertion 7.10.0. This process uses a fuzzy clustering toolbox to provide results for 5 clusters, but I didn't get a result for the clustering (closest hard clustering). I only got information about the cluster center and matrix iteration. So, what should I do to obtain that information? Any suggestions?
In the hard clustering, each samples allocated to only one cluster. This restriction is relaxed for fuzzy clustering and each sample can belong to more than one cluster with a certain degree of membership (Bezdek, 1981; Bezdek et al., 1992). This is particularly utile when the boundaries between clusters are indeterminate and not well separated. Moreover, the memberships may assist the users detect more sophisticated relationships between a given sample and the disclosed clusters. The fuzzy clustering can be changed to a hard clustering by assigning each sample to
the cluster with the largest measure of membership (Jain et al., 1999).
By doing this, a procedure should be implemented that is able to finf maximum membership value existing in U matrix (to find maximum membership value in each row of membership matrix).
Hi Novi, it's really easy!
I am supposing that you are calling the function like this:
[center,U] = fcm(data, 5);
In order to get an hard-clustering solution you just need to type:
[~,clustering_solution]=max(U)
The vector clustering_solution will contains for each element it's cluster number
The matlab function fcm provides two output parameters: the cluster centers and the partition matrix. To retrieve both parameters you need to write the function call as following: [centers,partition_matrix] = fcm(data, num_of_clusters). Then you can defuzzify the partition matrix using the max function.
Hi all,
Can you please let me know as to how one view the membership grade of each data point in matlab....I have a dataset containing three columns and 100 rows
- Badges
- Science topic