首页 | 本学科首页   官方微博 | 高级检索  
     


An intuitive clustering algorithm for spherical data with application to extrasolar planets
Authors:Wen-Liang Hung  Shou-Jen Chang-Chien  Miin-Shen Yang
Affiliation:1. Department of Applied Mathematics, National Hsinchu University of Education, Hsinchu, Taiwan;2. Department of Applied Mathematics, Chung Yuan Christian University, Chung-Li, Taiwan, Taiwan
Abstract:This paper proposes an intuitive clustering algorithm capable of automatically self-organizing data groups based on the original data structure. Comparisons between the propopsed algorithm and EM [1 A. Banerjee, I.S. Dhillon, J. Ghosh, and S. Sra, Clustering on the unit hypersphere using von Mises–Fisher distribution, J. Mach. Learn. Res. 6 (2005), pp. 139. [Google Scholar]] and spherical k-means [7 I.S. Dhillon and D.S. Modha, Concept decompositions for large sparse text data using clustering, Mach. Learn. 42 (2001), pp. 143175. doi: 10.1023/A:1007612920971[Crossref], [Web of Science ®] [Google Scholar]] algorithms are given. These numerical results show the effectiveness of the proposed algorithm, using the correct classification rate and the adjusted Rand index as evaluation criteria [5 J.-M. Chiou and P.-L. Li, Functional clustering and identifying substructures of longitudinal data, J. R. Statist. Soc. Ser. B. 69 (2007), pp. 679699. doi: 10.1111/j.1467-9868.2007.00605.x[Crossref] [Google Scholar],6 J.-M. Chiou and P.-L. Li, Correlation-based functional clustering via subspace projection, J. Am. Statist. Assoc. 103 (2008), pp. 16841692. doi: 10.1198/016214508000000814[Taylor &; Francis Online], [Web of Science ®] [Google Scholar]]. In 1995, Mayor and Queloz announced the detection of the first extrasolar planet (exoplanet) around a Sun-like star. Since then, observational efforts of astronomers have led to the detection of more than 1000 exoplanets. These discoveries may provide important information for understanding the formation and evolution of planetary systems. The proposed clustering algorithm is therefore used to study the data gathered on exoplanets. Two main implications are also suggested: (1) there are three major clusters, which correspond to the exoplanets in the regimes of disc, ongoing tidal and tidal interactions, respectively, and (2) the stellar metallicity does not play a key role in exoplanet migration.
Keywords:EM algorithm  extrasolar planets  mixtures of von mises distributions  spherical data  spherical k-means algorithm
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号