基于灰色模型的概念簇聚类算法

Conceptual clustering algorithm based on grey model

针对目前聚类算法的目标函数中尚未有能判别数据在时域上发展趋势差异性的情况,基于灰色理论中被广泛应用的预测模型GM(1,1),利用其所得到的预测公式,通过推导演算,设计新的聚类目标函数,构造了一种新的概念簇聚类算法,其聚类目标簇是在时域上具有共同发展趋势的数据集合。该算法在实际应用中的具体意义是:设计者可根据该算法的结果,对不同簇的数据设计不同的有针对性的预处理方案。以中国各城市人口发展统计数据为实例,验证了该算法的可行性。

Aimed at the current state that there is no object function in clustering algorithm,which can judge the differentiation of de-veloping tendency in time-area of data,based on the most widely used forecasting model in the grey theory GM（1,1）,using the forecasting formula of which,by speculating and computing,formed a new object function,then a new clustering algorithm for conceptual cluster is obtained,the object cluster of which is data sets having similar developing tendency.Based on the result,in realistic application,designers can make different effective dealing plan for different type of data.At last,vital statistics of cities in China is tested to validate the algorithm.