Research On Clustering Of LAMOST Stellar Spectra Based On Line Index

Clustering algorithm is an important algorithm used to find the data distribution and implicit scheme in data mining. It classifies large quantities of data points into several classes, minimizes the difference of the same classes, and maximizes the difference between the different classes.Line Index is a value representing some characteristics in the spectrum. Usually it is an integration of a certain period of magnitude of a spectrum, equivalent width of a spectral line (EW), or FWHM, it can also be a combination of some line index of a spectrum.This paper has summarized these applications of data mining methods in the survey data. This paper proposes a new method in response to these characteristics of the survey data:Clustering and Outlying Analysis of LAMOST Stellar Spectra Based on Lick Index.This paper finishes clustering of the survey data with k-means algorithm, using lick index as the eigenvalues of data with finished analysis results. The results show that the new method can gather data with similar physical characteristics together quicker and more efficient, with a very good results in the discovery of rare stars. This method can be applied to the study of Survey data. Details are as follows:(1) Review of the applications in the survey data of cluster analysis, outlier analysis and feature extraction. This paper summarizes the theory of cluster algorithms with outlier analysis and their classifications, reviews their applications in astronomy, and sums up the applications in the survey data of PCA and index.(2) Researches on clustering of LAMOST stellar spectra is based on Lick index. This study finishes clustering of the survey data by k-means algorithm, using lick index as the eigenvalues of data. The results show that the new method can gather data with similar physical characteristics together quicker and more efficient, and there are significant dissimilarities in between clusters.(3) Researches on outlier analysis of LAMOST stellar spectra is based on Lick index. This study analyzes the clusters which are smaller or mean spectrum is special so that emission-line star, aging M star, metal-poor star, and etc. could be found. The study shows that the method of clustering of LAMOST stellar spectra based on Lick index has good results in finding the rare stars.To sum up, clustering and outlier analysis of LAMOST stellar spectra based on index can lead to good results. They can gather data with similar physical characteristics together and find the rare data quicker and more efficient. Algorithm models in this paper need further research to achieve better clustering performances, so that it can be applied to the LAMOST.