Multi-objective data stream clustering
Mohammed Oualid Attaoui, Hanene Azzag, Mustapha Lebbah, Nabil Keskes · 2020
A Data stream is a massive sequences of data coming continuously. Clustering this type of data requires some restrictions in time and memory. Most of the clustering algorithms follow only one cluster validity measure. Given different data properties, a single validity measure does not work well for all datasets. In this paper, we introduce MOC-Stream based on Multi-objective clustering and data stream concepts. The goal of MOC-Stream is to find clusters by applying several algorithms corresponding to several objective functions. It uses a two-phase process: 1) online phase: creating several clustering solutions based on different algorithms and genetic operators 2) offline phase: building an optimal partition from the discovered clusters. Experiments on large stream datasets show the effectiveness of MOC-Stream for detecting arbitrary shaped, compact, and well-separated clusters with better execution time.