OPTICS (ordering points to identify the clustering structure) is a density-based clustering algorithm. OPTICS is capable of finding non-circular clusters, unlike other algorithms like DBSCAN and k-means.

This visualization aims to help you understand how OPTICS works. The chart on the upper left estimates the densities of the input data. Doubleclicking on it will toggle the actual data points. These data points can be selected using a lasso. Doing so will highlight them in the chart below, the reachability chart.

On the upper right, a small bar chart showing the cluster sizes can be found. Below that, we've plotted all the jumps that the algorithm made during execution. Hovering a bar of the reachability chart (lower left) will highlight the corresponding edge in this chart, as well as enlarge the corresponding point in the density chart and show its epsilon neighborhood.

The heatmap on the lower right shows the distances between two points i and j. The points are ordered by the algorithm and mirrored on both axes. This view makes it easier to find clusters and even subclusters---although it shows the same data as the reachability chart, the square shapes are much easier to see for humans.

OPTICS is a cool and robust algorithm. We encourage you to check out the data tab below to select another data set, or even enter your own, and the settings tab below that to change the parameters of the algorithm and the visualization.