| 摘要: |
| 大数据时代下收集到的数据常含有异常值或呈现尖峰厚尾以及变量之间具有较强的相关性，针对此问题，结合秩回归和自适应弹性网(Adaptive Elastic-net )提出了一种高效稳健的变量选择方法。此方法的最大优点在于不仅能够有效处理协变量之间的强相关性而且还能克服多重共线性问题，同时能抵抗厚尾分布或异常值的影响，实现稳健的变量选择。在数值计算方面，采用二次近似和牛顿迭代算法以获得新变量选择方法的稳定数值解，仿真实验表明：新提出的方法比现有方法表现更好，特别是对于厚尾分布或异常值的情况。最后，通过对中国重要的股票市场指数——中证100指数的跟踪，进一步表明该方法在有效样本下具有良好的表现。 |
| 关键词: 多准则决策 最小叉熵 垂面距离 主客观权重集成 创新项目优选 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| Hybrid Multi-criteria Decision Making Method and Its Application to Innovative Project Optimization in Pilot Free Trade Zone |
|
HAN Er-dong
|
|
School of Business, Luoyang Normal University, Henan Luoyang 471934, China
|
| Abstract: |
| In order to deal with the mixed multi-criteria decision making problem with unknown criteria weight, a hybrid multi-criteria decision making method based on minimum cross entropy and vertical distance is proposed. The mixed decision information is normalized and processed by the normalized methods of different types of evaluation information, the objective weight of each criterion is determined by the information entropy method, single-valued utility function reflects the subjective judgment of each decision maker on the importance of different criteria, and the subjective weight of each criterion of each decision maker is obtained. Based on the minimum cross entropy, the subjective combination weight of each criterion is obtained by integrating the subjective weights vector of each criterion. An optimization model is constructed to minimize the sum of the cross entropy between criterion’s comprehensive weight and subjective weight vector, and objective weight vector. Thus, the comprehensive weight vector of each criterion is determined. The ranking of the alternatives is obtained by the vertical distance between the alternatives and the positive or negative ideal solutions. Finally, the hybrid decision- making method is applied to the optimization of innovative projects in Henan Pilot Free Trade Zone, and the validity and feasibility of the decision making method are verified by comparative analysis. |
| Key words: multi-criterion decision making minimum cross-entropy vertical distance subjective-objective weight integration innovation project optimization selection |
