A Fast K-Means Clustering Algorithm Based on Grid Data Reduction

K-Means is a popular clustering algorithm to find the clustering easily by iteration. But the computational complexity of the traditional k-means due to accessing the whole data in each cycle of iterative operations is too great to make it fit for very large data set. This paper presents a new clustering algorithm we have developed, fast k-means clustering algorithm based on grid data reduction (GDR-FKM), by which clustering operations can be quickly performed on very large data set. Application of the algorithm to analysis of the data relativity in TT&C has demonstrated its celerity and accuracy.