目录
图像
上传图像
DSS Images Other Images
相关文章
Random forest automated supervised classification of Hipparcos periodic variable stars
We present an evaluation of the performance of an automatedclassification of the Hipparcos periodic variable stars into 26 types.The sub-sample with the most reliable variability types available in theliterature is used to train supervised algorithms to characterize thetype dependencies on a number of attributes. The most useful attributesevaluated with the random forest methodology include, in decreasingorder of importance, the period, the amplitude, the V-I colour index,the absolute magnitude, the residual around the folded light-curvemodel, the magnitude distribution skewness and the amplitude of thesecond harmonic of the Fourier series model relative to that of thefundamental frequency. Random forests and a multi-stage scheme involvingBayesian network and Gaussian mixture methods lead to statisticallyequivalent results. In standard 10-fold cross-validation (CV)experiments, the rate of correct classification is between 90 and 100per cent, depending on the variability type. The main mis-classificationcases, up to a rate of about 10 per cent, arise due to confusion betweenSPB and ACV blue variables and between eclipsing binaries, ellipsoidalvariables and other variability types. Our training set and thepredicted types for the other Hipparcos periodic stars are availableonline.
| The 74th Special Name-list of Variable Stars
We present the Name-list introducing GCVS names for 3153 variable starsdiscovered by the Hipparcos mission.
|
提交文章
相关链接
提交链接
下列团体成员
|
观测天体数据
目录:
|
