DBSCAN Clustering

The DBSCAN Clustering algorithm provides a way to group together similar data points. This similarity can be measured using any metric but Euclidean Distance is most commonly used.

The algorithm works roughly as follows:

There is a cluster when the number of points within distance ε is greater than some min points threshold. We go through each point and build clusters by finding points densely connected to it.

Interactive DBSCAN clustering.

The above lets you test out a DBSCAN Classifier. Click to add new data points and hit 'Cluster' to run the classifier.