Affiliations: [a] School of Management Science and Information Engineering, Jilin University of Finance and Economics, Changchun, China
| [b] School of Management, Jilin University, Changchun, China
| [c] Department of Computer Science, State University of New York, New Paltz, NY, USA
Correspondence:
[*]
Corresponding author. Keqin Li, Department of Computer Science, State University of New York, New Paltz, NY 12561, USA. E-mail: lik@newpaltz.edu.
Abstract: The density peaks clustering (DPC) algorithm is a novel density-based clustering approach. Outliers can be spotted and excluded automatically, and clusters can be found regardless of the shape and of dimensionality of the space in which they are embedded. However, it still has problems when processing a complex data set with irregular shapes and varying densities to get a good clustering result with anomaly detection. A density fragment clustering (DFC) algorithm without peaks algorithm is proposed with inspiration from DPC, DBSCAN and SCAN to cope with a larger number of data sets. Experimental results show that our algorithm is more feasible and effective when compared to DPC, AP and DBSCAN algorithms.
Keywords: Density peak, fragment clustering, anomaly detection
