Hierarchical clustering algorithm is a well-known algorithm for clustering data points into different clusters. The number of clusters may not be specified apriori unlike K-means. Although, a cutting point value can be used to obtain clusters at a particular point during the clustering operation. The algorithm is either bottom-up (agglomerative) or top-down (divisive).
Recently I needed to cluster large dataset using HAC algorithm, so I checked Mahout list of clustering algorithms. Mahout has limited algorithm for clustering operation and all of these require the knowledge of cluster k. I tried HAC implementation on WEKA but the algorithm run for two days with 2GB heap memory size without producing any results. Any advice on this? Thanks
I believe that for massive data the easiest variant of hierarchical clustering can be taken from F.MURTAGH and Contreras many articles. Fast linear time...
IT is easy to use Baire distance for hierarchical data representation. Just check GOOGLE scholar...
Dear Adewole,
You can find an old discussion on the topic at StackOverflow that includes some relevant references:
http://stackoverflow.com/questions/84820/distributed-hierarchical-clustering
Best regards,
-- Fernando
Distributed Hierarchical Clustering's following link may help http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=5685388&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D5685388
@Leo Vohandu,@Fernando Berzal, and @Harsh Bhasin thank you for your suggestions. I will do as you have suggested. I will get back to you on this. Thanks.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining