Skip to content
alitouka edited this page Nov 24, 2014 · 10 revisions

Spark DBSCAN is an implementation of the DBSCAN clustering algorithm on top of Apache Spark . It also includes 2 simple tools which will help you choose parameters of the DBSCAN algorithm.

Clusters identified by the DBSCAN algorithm

This software is EXPERIMENTAL , it supports only Euclidean and Manhattan distance measures ( why? ) and it is not well optimized yet. I tested it only on small datasets (millions of records with 2 features in each record).

You can use Spark DBSCAN as a standalone application which you can submit to a Spark cluster ( Learn how ). Alternatively, you can include it into your own app - its API is documented and easy to use ( Learn how ).

Learn more about:

Performance

Performance chart

Credits

I was glad to receive contributions from other people and I'd like to say thank you:

  • Mark Geraty - for fixing a bug with Java RDDs;
  • @agrinh - for adding compatibility with Spark 1.1.0

Contact me

Any questions, comments, suggestions as well as criticism are welcome! You can contact me via: