| With the development of information technology, data mining technology causes the attention of people, especially for astronomy neighborhood of the need to use large amounts of data. The development of data mining technology promoted the progress of each neighborhood.AutoClass C(automatic) clustering algorithm is an unsupervised Bayesian classification system, it can be divided into different classes of the data automatically,and record the probability.of each data respectively belong to different kinds. We can through the classification results to determine whether classified accuracy achieve our desired outcomes for data classification.In this paper, we use AutoClass C to classify regular graphics,such as oval and triangle,and then classify known globular clusters. We use three different kinds of data attributes and three different density ratio(the ratio of the members of the star and the background stars) to test. For the two data attributes combination, the classification of different density ratio, density ratio is the greater then the error is smaller, the members of the cluster member account for the total ratio is almost the same stars, but there are some special cases, such as location x and y as data attributes,the classification of number density ratio of 10:1. For multiple data attributes combination, the classification of different density ratio, density ratio is the greater then the error is smaller, the members of the cluster member account for the total ratio is almost the same, and the ratio of almost 95% or more; and it is Compared with the classification of the two data attributes to participate in, under the condition of the same number density ratio, the error is very small, and the star cluster of the total members of the star ratio is higher, the overall classification ratio is higher.Therefore, we can apply AutoClass C to find the possibility of unknown globular clusters in the region in the future, and bring the convenience for astronomers in their last work. |
PDF Full Text Request