用AutoClass来寻找银河系中的球状星团

发布时间：2017-12-27 12:43

本文关键词：用AutoClass来寻找银河系中的球状星团　出处：《上海师范大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着信息技术的发展,数据挖掘技术越来越引起人们的关注,尤其是对于需要运用大量数据的天文学邻域。数据挖掘技术的发展推动了各邻域的技术进步。AutoClass C(自动聚类算法)是一种无监督的贝叶斯分类系统,它可以将数据自动的分为不同的类,并记录每一条数据分别属于不同类的概率。我们可以通过分类结果,对分类正确率进行判定是否达到我们对于数据分类的预想结果。本文我们先用AutoClass C对规则图形椭圆和三角形进行分类测试,其次再对已知球状星团进行分类测试。我们采用三种不同的数据属性组合,选用不同的密度比(成员星与背景星之比)进行测试对比。对于两个数据属性组合、不同密度比的分类,密度比越大,误差越小,得到的星团成员星占总的成员星比率几乎不变,但也有一些特殊情况,比如以位置x和y为数据属性,数密度比为10:1的分类。对于多个数据属性组合、不同密度比的分类,同样密度比越大,误差越小,得到的星团成员星占总的成员星比率几乎不变,且比率几乎都在95%以上;与两个数据属性参与的分类相比,在数密度比相同的情况下,误差很小,且得到的星团成员星占总的成员星比率更高,总体分类效果更高。因此,未来我们可以应用AutoClass C来寻找天区中的未知球状星团的可能性很大,给天文学家以后的工作来带了方便。
[Abstract]:With the development of information technology, data mining technology has attracted more and more people's attention, especially for the astronomical neighborhood which needs to use a large amount of data. The development of data mining technology has promoted the technological progress of the neighbourhood. AutoClass C (automatic clustering algorithm) is an unsupervised Bias classification system. It can automatically divide data into different classes, and record the probability that each data belongs to different classes. We can determine whether the classification accuracy is up to the expected results of our data classification by classifying the results. In this paper, we first used AutoClass C to test the ellipse and triangle of regular pattern, and then classify the known globular clusters. We use three different combinations of data attributes and use different density ratios (the ratio of member stars to background stars) to compare the tests. For the classification of different density than the two data attribute combination, the density ratio increases, the error is small, the total members of the star cluster member of astrology ratio is almost the same, but there are also some special cases, such as the location of X and y for the data attribute, number density ratio of 10:1 classification. For the classification of different density than the multiple attribute data combination, the same, the greater the density ratio, the error is small, the total members of the star cluster member of astrology ratio is almost the same, and the ratio of almost 95%; compared with the classification of the two data attributes involved, in the number density ratio under the same error is very small, and the total members of the star cluster member astrology rate higher, higher overall classification effect. Therefore, the possibility of future we can use AutoClass C to find the unknown area in large globular clusters, give astronomers future work to bring convenience.
【学位授予单位】：上海师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P156

【相似文献】