| Cui Yanzhe,Zhao Lindan.Unbiased Weighting Method Based on Cross Entropy[J].The Journal of quantitative and technical economics,2020,(3):181-195 |
| 基于交叉熵的无偏赋权法 |
| Unbiased Weighting Method Based on Cross Entropy |
| |
| DOI： |
| 中文关键词: 熵权法信息熵交叉熵多项Logit回归 |
| 英文关键词: Entropy Weight Method Information Entropy Cross Entropy Multiple Logit Regression |
| 基金项目: |
|
| Hits: |
| Download times: |
| 中文摘要: |
| 研究目标：提出一种基于交叉熵的无偏赋权法。研究方法：在熵框架下，借鉴机器学习中的分类算法和计量经济学中多项Logit回归的相关研究方法，对信息熵熵权法进行无偏优化，提出了交叉熵熵权法，并且结合数值模拟和实例考察的方法对比信息熵熵权法和交叉熵熵权法的赋权效果。研究发现：交叉熵熵权法从理论基础和数值模拟两个方面均证明了权重的无偏性和稳健性，交叉熵熵权法和信息熵熵权法具有相同的适用性，权重具有相同水平的保性序，但是交叉熵熵权法的权重更加契合前人的研究成果，合成结果具有更大方差。研究创新：从熵、机器学习和计量经济学三个角度论证了交叉熵熵权法，保证了赋权结果的无偏性和赋权方法的可操作性。研究价值：构建的赋权方法为多指标评价体系的研究提供了强有力的分析工具。 |
| 英文摘要: |
| Research Objectives: To propose an unbiased weighting method based on cross entropy.Research Methods: In the entropy framework, based on the classification algorithm in machine learning and the related research methods of multiple Logit regression in econometrics, the information entropy weight method is optimized unbiasedly, and the cross entropy weight method is proposed.The evaluation results of entropy weight method and cross entropy weight method are compared by numerical simulation and case study.Research Findings: The cross entropy weight method prove the unbiasedness and robustness of weights from both theoretical and numerical simulations.Cross entropy weight method and information entropy weight method have the same applicability, weight has the same level of conservation order, but the weight of cross entropy weight method is more consistent with the previous research results, the composite results have greater variance.Research Innovations: The cross entropy weight method is demonstrated from the perspectives of entropy, machine learning and econometrics, which guarantee the unbiasedness of the weighting result and the operability of the weighting method.Research Value: The constructed weighting method provides a powerful analytical tool for the research of multi-index evaluation system. |
|
View Full Text
View/Add Comment Download reader |
| Close |
