Focused on the disadvantage of classical Euclidian distance in data clustering analysis, we propose an improved distance calculation formula, which describes the local compactness and global connectivity between data points. Furthermore, we i