Research Of Grid-based Clustering Algorithm

The research on grid-based clustering algorithms have become a highly active topic in the domain of data mining research. Based on the analysis of traditional grid-based clustering algorithms, we bright forward the MST Clustering Algorithm Based on Optimized Grid (OGMST). It resolves the GMST(Grid-based MST Clustering Algorithm) algorithm's limitation of unfit for multi-density datasets by the use of parameter automatic density threshold method. Besides, the OGMST algorithm disposes the border points by the method of border points disposed technique that enhances remarkably the precision of grid-based clustering. Aiming at the limitations of traditional measurement method on similitude between objects, we put forward Grid-Similarity-based Clustering Algorithm (GSCA), it brings in a new criterion to measure the similitude between objects. It applies on the grid clustering and disposes the density threshold of grid by the method of density threshold that improves the precision of clustering. Besides, the GSCA algorithm disposes the very high dimension datasets by the technique of entropy. The two algorithms show their advantages in the series of comparative experiments with some traditional clustering algorithms.