基于R型聚类-因子分析的指标体系简化方法

摘要

An excessively complicated index system may increase the complexity of computation. This work proposes a simplification method of index system based on R cluster analysis and factor analysis to extract the representative elements. This method can control the information loss rate of the representation element and explain the actual meaning of the representation element easily. It also avoids the weakness of factor analysis which is difficult to handle the problem with independent factor. The experiment results show that this method can improve the simplicity and effectiveness of index system significantly, and maintain the completeness of the index information without a prior knowledge is available.%评价指标体系过于复杂，容易产生冗余信息，增加计算与分析的难度。本文针对评价指标体系的简化方法展开研究，提出了基于 R 型聚类-因子分析的代表元提取方法。该方法具有定量控制代表元的信息丢失率、代表元实际含义易于解释等优点，同时避免了因子分析难以处理独立指标的问题。实验结果表明本文所提出的R型聚类–因子分析指标体系简化方法可以在没有先验知识可用的情况下，有效提高指标体系的简明性与有效性，同时保持指标信息的完备性。