一种面向高维数据的DS-ALasso变量选择方法
首发时间:2018-07-02
摘要:变量选择是高维数据分析的重要环节,Laaso方法不具有Oracle性质且存在在处理维数大于样本量的数据结构时只能选取个变量的缺点,为解决此问题,提出一种改进的Lasso方法:DS-ALasso方法,该方法是Dantzig Selector和Adaptive Lasso两种方法的结合而成的一种能把高维数据的维数降低到小于样本量并具有Oracle性质的有效方法。实验表明,均分式Adaptive Lasso方法能够很好地对高维海量或高维小样本数据集进行特征选择,是一种有效的特征选择方法。
For information in English, please click here
High-dimensional data feature selection method based on DS-ALasso
Abstract: Variable selection is of great importance in High-dimensional data analysis.The Laaso method does not have the Oracle property and can only select the disadvantages of a variable when dealing with the data structure with the dimension larger than the sample size.To address this issue, this paper proposes an improved Lasso method, called DS-ALasso. The combination of Dantzig Selector and Adaptive Lasso can be an effective method which can reduce the dimensionality of high-dimensional data to less than the sample size and have the property of Oracle. Experimental results show that the DS-ALasso.as an effective feature selection method. can effectively deal with the high-dimensional and large sample datasets.
Keywords: Lasso variable selection DS-ALass.
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
一种面向高维数据的DS-ALasso变量选择方法
评论
全部评论0/1000