There are plenty of publications that propose methods to choose the parameters eps and minpts in DBSCAN. The most notable is OPTICS, a DBSCAN variation that does away with the epsilon parameter; it produces a hierarchical result that can roughly be seen as "running DBSCAN with every possible epsilon". but still in weka when i use OPTICS i have to provide the minpts and epsilon parameters.
Is it a different implementation of OPTICS?
You are very correct with your explanations, however to be frank with you, OPTICS algorithm implementation on WEKA is very basic, slow, incomplete and unmaintained. I didn't know what you are trying to solve, but I must tell you there are other interesting clustering algorithms on ELKI machine learning tool. You can join this discussion (http://stackoverflow.com/questions/10469287/clustering-interacting-particles-with-weka) to find out more. I believe there will be useful tips for you. All the best.
Hi Iftikhar ,
In WEKA, DBSCAN applies (as a function) Nearest-neighbor-based clustering, that automatically decides the number of clusters. Precisely, it uses the Euclidean distance metric to determine which instances belong together in a cluster, while OPTICS is the extension of DBSCAN to hierarchical clustering.
HTH.
Samer
- Badges
- Science topic
- Similar topics
- Data Mining
- Classification