Reputation: 21
Clustering techniques for similarity matrix
I have a binary data of 128 respondants based on the features of digital camera that they have selected. where '1' represents the selection of feature and '0' represents that feature not selected. i have 92 product features in columns and respondants in rows. Each respondant has exactly selected 20 features out of set of 92 features. I want to create the clusters of different user groups based on the features they selected. I have tried some clustering algorithms like fuzzy clustering and hierarichal on these binaray data but it didnt gave me any good results and the clusters created were really bad. So now i have applied the dice coefficient similarity matrix on the data w.r.t the respondants, that basically gives me the similarity score for each respondant with all the other respondants. Is it possible to apply clustering technique on this similarity matrix to get good clusters? also what clustering techniques are available that i could apply on this user similarity matrix so that i could identify the clusters of users based on their similairty score. Any suggestion and comment would be really appreciated
Upvotes: 0
Views: 440
Answers (1)
Reputation: 77454
Since your data set is tiny, go with hierarchical clustering.
It can be implemented with distance or with similarity.
Upvotes: 0
Related Questions
- Similarity matrix clustering
- Clustering techniques for Binary Data
- Agglomerative hierarchical clustering technique
- Clustering given a distance matrix 256x256
- How to do Hierarchical Clustering for large similarity matrix
- Complete link clustering
- R data clustering using a pre-defined distance/similarity matrix
- Effective clustering of a similarity matrix
- Hierarchical Cluster Analysis in Cluster 3.0
- Hierarchical Clustering Large Sparse Distance Matrix R