摘要
分级界限的确定一直都是学者们关注的焦点,而如何确定分级数则往往被忽略了。根据模糊覆盖和膨胀因子的定义,提出了一种确定1维数据分级数的分级方法:首先通过覆盖半径做一个模糊覆盖,利用自膨胀因子对这个覆盖进行扩张,直到不能再向外扩张为止,这一级就确定了;然后进行下一级的确定。该方法不仅可以自动确定数据的分级数,而且克服了分级结果对参数的敏感性。最后用实验证明了方法的有效性。
Determining the clustering limit was a focus of scholars＇attention, but how to ascertain the number of clustering was often ignored. This paper put forward a new clustering method for one dimension data, according to the definitions of fuzzy cover and adaptive inflation factor. First, we used cover-radius to make a fuzzy-cover, expanded it by adaptive inflation factor until it could not be expanded, and then ascertained the next cluster. This approach did not only automatically make sure of the number of clustering, but also solved its impressible to parameter. Finally, the experiments verify the validity of the approach.
出处
《测绘科学技术学报》
北大核心
2007年第5期384-386,共3页
Journal of Geomatics Science and Technology
基金
国家自然科学基金资助项目(40501058
40401052)
关键词
分级数
模糊覆盖
膨胀因子
有效性
number of clustering
fuzzy cover
inflation factor
validity
