evalcluster for fuzzy c-means ?

2 次查看(过去 30 天)
Vincent Moron
Vincent Moron 2023-7-5
回答: Prathamesh 2023-7-27
Dear all
I wonder if several score used to evaluate hard clusters (Silhouette, Gap, Davies-Bouldin, etc.) works also for fuzzy clusters ? For fuzzy clusters, the "optimal" number should include (1) the value of the fuzzifier (=m) and (2) the number of clusters (=n). Any idea (or code) using both informations and leading to a unique score for values of m and n will be useful.
Thank you in advance
Vincent M.

回答(1 个)

Prathamesh
Prathamesh 2023-7-27
Hi,
I understand that you want to know how to get unique score matrix for the values of fuzzifier and number of clusters.
You can use fcm function from fuzzy logic toolbox to get the membership matrix ‘U’.
options = fcmOptions('NumClusters', 3, 'Exponent', 2);
[centers, U] = fcm(x, 3, options);
U is a matrix that contains the membership grades of each data point to each cluster. Each row of U corresponds to a data point, and each column corresponds to a cluster.
U transpose will be the score matrix or particular value of fuzzifier and number of clusters.
The “evalclusters” function can be used to find the optimum number of data clusters.
eva = evalclusters(x, U, criterion);
Refer to the documentation of the above function to get more information.
I hope this resolves your issue.

类别

Help CenterFile Exchange 中查找有关 Data Clustering 的更多信息

产品


版本

R2023a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by