基于Logistic回归模型的Blazar天体的分类

从文献中收集了205个Blazar天体,包括142个BL Lac天体和63个平谱射电类星体(FSRQs).对这些天体的类别与它们的红移、射电5 GHz辐射流量、光学V波段流量、1 keV处X射线流量、X射线光子谱指数进行了相关性和Logistic回归分析.结果表明,对Blazar天体分类产生主要影响的因素是红移、射电5 GHz辐射流量和X射线光子谱指数,综合应用这三个物理量判别Blazar天体的分类的准确率可达到91.2%,得到的分类方程具有良好的预测效果,可以作为Blazar天体分类的一个重要的判据.而光学V波段流量和1 keV处X射线流量不能区分开BL Lac天体和FSRQs,它们与Blazar天体分类没有相关性.本文结果支持将BL Lac天体和FSRQs归为Blazar天体,不同类别的Blazar天体之间能通过一种演化序列相联系.关键词：Logistic分类Blazar天体

Classification of Blazar objects based on Logistic regression model

We collecte 205 Blazars (142 BL Lacertae objects and 63 flat spectrum radio quasars(FSRQs)),and investigate the correlation between the redshift, the fluxes in the radio band (5 GHz), optical (V band) , X-ray (1 keV), and the X-ray photon spectral index for the BL Lacertae objects and FSRQs by the correlation analysis and Logistic regression model. The results indicate that major influencing factors on Blazar object classification are the redshift, the fluxes in the radio band (5 GHz), and the X-ray photon spectral index. The derived equation is effective as a substantial criterion for Blazar objects classification, of which the accuracy is 91.2%. Our results strongly support the division of Blazar objects into BL Lacertae objects and FSRQs. We compare the properties of BL Lacertae objects and FSRQs from the fluxes in the optical (V band) and X-ray (1 keV), and find that BL Lacertae objects are not significantly different from FSRQs. The implications of these correlations, which support the evolutionary sequence of Blazar objects, are also discussed.