采用机器学习的聚类模型特征选择方法比较

Comparison of Feature Selection Method of Clustering Model Using Machine Learning

针对机器学习聚类模型在特征选择时存在的问题,首先,对特征选择在聚类模型中的适用性进行分析并对其进行调整和改进.然后,基于R语言中的递归特征消除(RFE)特征选择方法和Boruta特征选择方法进行特征选择算法设计.最后,应用聚类内部有效性指标,对在线品牌忠诚度聚类模型优化结果进行分析,进而对特征选择方法进行比较研究.结果表明:Boruta特征选择方法更具优势.

Targeting at problems during the feature selection process of machine learning clustering model, at first,it makes analysis on the applicability of the feature selection for clustering model and makes adjustment and improvement. Then makes feature selection algorithm design based on R language recursive feature elimi-nation （RFE） feature selection method and Boruta feature selection method. At last, applying cluster interior validity indexes to analyze the optimization result of online brand loyalty clustering model, a further compara-tive study is made on the feature selection method. The results show that the Boruta feature selection method has more advantages.