关键词: k-means聚类, 粗糙集, 指标筛选, 指标筛选, k-means聚类, 粗糙集

Abstract: Aiming at the information duplication and uncertainty of indicators in multi-attribute decision, an indicators screening method is proposed based on improved k-means clustering and rough set algorithm. Firstly, we define the spatial distribution density of the sample, and implement the k-means algorithm of the initial cluster center optimization, so the continuous indicators are discretized. Then we use the relative reduction principle to reduce the indicators, and delete the redundant indicator with repeated information. Combined with the case of green economic indicator system, the rationality and effectiveness of the method are verified.

Key words: k-means clustering, rough set, indicators screening, Indicators screening, k-means clustering, rough set