综合评价中确定权重向量的几种方法比较

Comparing Several Methods of Assuring Weight Vector in Synthetical Evaluation

以企业技术创新能力评价排序问题为实证范例，对德尔菲法、层次分析法、熵值法和模糊聚类分析法测定评价指标权重的可信度做了比较研究，目的是探索各自的适用范围和优缺点．文献通过理论分析特别是实证比较，认为样本数据完整，熵值法则较高；如果缺乏样本数据，则应采用AHP法，可信度高于德尔菲法；若样本中含有大量的模糊数据，且同一层次指标个数较多，应先采用模糊聚类分析法分类，再用AHP法分别做各个子类的权重．

：Based on the evaluation and sorting of enterprise's technolo gy innovation ability, this paper compares the reliability of measurin g weight of evaluating index by Delphi method, Analytic Hierarchy Proc ess(AHP), Entropy method and Fuzzy Cluster Analysis, to research appli cable area and advantage and disadvantage of each method. The paper dr aws conclusion through theoretical analysis especially practical compa re that if complete sample datum can be available, the reliability of Entropy method is higher. If sample datum is scare, the reliability of AHP is higher than Delphi method. If there are a lot of fuzzy indexes among sample, we should adopt Fuzzy Cluster Analysis, then conclude t he weights of indexes by AHP.