Is this possible to plot Subtractive clustering?

Dear all,

I am new to the clustering and machine learning plaza. I have used 'subclust' to cluster a dataset using the subtractive method. Now, apparently the algorithm only gives me the centers of each cluster. I was wondering if I can plot the clustered dataset?

(using C-Means 'fcm' it automatically generates the fuzzy partition matrix (U) which and finally enables you to plot the clusteded dataset. but apparently such option is not available with subtractive. Am I right?? If so, then what is the point of subtractuve clustering? only finding the center of clusters?

[Image Analyst]

How many clusters do you expect? Can you show a scatterplot of the data? I've not heard of subtractive clustering. Why did you choose that particular method over any of the others, like kmeans, knn, or dbscan?

[Sam Chak]

Hi @Alireza Babaei

Approah #1

If your intention is to plot or visualize only the clusters, then you can use the Clustering Tool.

https://www.mathworks.com/help/fuzzy/data-clustering-using-the-clustering-tool.html

Approah #2

The accuracy of the clustering result using the fuzzy c-means (fcm) method generally depends on initial cluster number. By using a hybrid fuzzy clustering scheme, you can first use the subtractive clustering method (subclust) for estimating the number of centers of clusters, and then use the fcm algorithm to generate the fuzzy partition matrix.

Approah #3

The function subclust can also returns the sigma (σ) value that indicates the range of influence of a cluster center. Having the center and sigma values, theoretically, you should be able to construct the fuzzy partition matrix.