Suppose, we bicluster a dataset containing data in row and column fashion about an event. After biclustering, what can we say by interpreting the clustering result?
.
say you have a set of documents represented in a bag of words fashion
bi-clustering will end up with clusters of documents so that documents in a cluster are distributed in the same way across the clusters of words and that such distributions are different from one cluster to the other (and the same thing holds regarding the clusters of words and their distribution across clusters of documents)
you have built two clusterings (one for the documents and one for the words) which have maximum mutual information with each other
.
Unlike K-means, Bi-clustering algorithm makes clusters for rows and columns separately and try to find optimized clusters among them.
You can use a famous software toolbox "BiCAT"......where you need to fed your dataset and you can use any algorithms from the toolbox of it.....
You may get some help if you follow the below links,
Biclustering: Algorithms and Application in Data Mining by Petros Xanthopoulos, Nikita Boyko, Neng Fan, Panos M Pardalos(Wiley)
https://en.wikipedia.org/wiki/Biclustering
I think Fabrice has given an excellent example, not in all case bi clustering may be significant.
- Badges
- Science topic
- Similar topics
- Books