基于灰关联度的面板数据聚类方法及在空气污染分析中的应用

Panel data clustering method based on grey correlation degree and its application in air pollution analysis

针对面板数据的三维特征、正负相关性以及灰色关联模型缺乏有效检验等问题,构建面板数据灰色关联模型,拓展了面板数据聚类方法;引入灰熵,提出灰色关联聚类的检验模型.研究发现,灰色关联度模型能够反映面板数据的正、负相关关系,且具有对称性、唯一性和可比性.通过在苏南4市空气质量区域划分中的应用,表明基于面板数据灰关联模型的聚类方法具有良好的效果,各类别的灰色关联度差异明显,层次清晰.

According to some problems of the grey relational model for panel data, such as the three-dimensional feature,the positive and negative relation of the panel data, and the model being lack of effective testing, the panel data grey relational model is built to expand the panel data clustering method. Grey entropy is introduced into the model and the test model of grey clustering model is proposed. The study shows that the grey relational model can reflect the positive and negative relation of the panel data and possess symmetry property, unique property and comparable property. The application of example in air pollution analysis indicates that the method has good effect, and the grey correlation degree of each category is obvious.