A boundary restricted adaptive particle swarm optimization for data clustering

Data clustering is the most popular data analysis method in data mining. It is the method that parts the data object to meaningful groups. It has been applied into many areas such as image processing, pattern recognition and machine learning where the data sets are of many shapes and sizes. The most popular K-means and other classical algorithms suffer from drawback of their initial choice of centroid selection and local optima. This paper presents a new improved algorithm named as Boundary Restricted Adaptive Particle Swam Optimization (BR-APSO) algorithm with boundary restriction strategy. The proposed BR-APSO algorithm is tested on nine data sets, and its results are compared with those of PSO, NM-PSO, K-PSO and K-means clustering algorithms. It has been found that the proposed algorithm is robust, generates more accurate results and its convergence speed is also fast compared to other algorithms.