# CRAD: Clustering with Robust Autocuts and Depth

## TL;DR

CRAD is a novel density-based clustering algorithm utilizing robust data depth for neighbor searching, effectively detecting clusters with varying densities and extending to time series data with an automatic parameter selection.

## Contribution

Introduces CRAD, a new clustering method based on data depth, with a parameter selection procedure and extension to time series clustering.

## Key findings

- CRAD outperforms DBSCAN, OPTICS, and DBCA in detecting clusters with varying densities.
- The parameter selection procedure effectively identifies optimal parameters in real-world data.
- CRAD successfully extends to time series clustering without prior knowledge of cluster count.

## Abstract

We develop a new density-based clustering algorithm named CRAD which is based on a new neighbor searching function with a robust data depth as the dissimilarity measure. Our experiments prove that the new CRAD is highly competitive at detecting clusters with varying densities, compared with the existing algorithms such as DBSCAN, OPTICS and DBCA. Furthermore, a new effective parameter selection procedure is developed to select the optimal underlying parameter in the real-world clustering, when the ground truth is unknown. Lastly, we suggest a new clustering framework that extends CRAD from spatial data clustering to time series clustering without a-priori knowledge of the true number of clusters. The performance of CRAD is evaluated through extensive experimental studies.