基于支持向量机的多属性决策方法

发布时间：2018-08-15 16:07

【摘要】：多属性决策也称有限多方案多目标决策,即考虑多个方案,且每个方案有多个属性进行描述,这类问题在很多领域都很常见,因此该研究具有深刻的理论意义和广泛的实际应用背景。对于多属性决策问题,往往希望能够绝对客观公正地给决策者最优的备选方案,正是基于这种考虑,本文提出了利用支持向量机的原理,对决策问题进行回归分析拟合,从而得出决策模型的机制。支持向量机是一种基于统计学习理论的机器学习方法,它具有较为完备的理论基础和较好的学习性能。目前统计学习理论正处于一个向实际应用推广的阶段,支持向量机需要进一步完善和改进,以满足实际应用的需求。支持向量机尤其对于小样本、非线性问题展现了较为良好的性能。所以本文利用了支持向量机在多属性决策问题上的适应性,给出了基于支持向量机的多属性决策方法。本文首先研究了支持向量回归模型的参数问题,为了使得模型参数的选取更为迅速,引入了粒子群算法。将该模型应用到一类多属性决策问题中,对该问题进行回归拟合,并给出决策方案,通过对比实验看出优化模型的性能。接着对多属性决策中的期刊评价问题进行实验分析。由于期刊评价问题的属性间具有较强的相关性,这就使得一些回归拟合方法具有限制性,而运用支持向量回归的方法可以不用考虑以上问题。最后,对属性更多的更为复杂的多属性决策问题进行研究,引入主成分分析算法对这类问题进行简化,最大程度地去除冗余和不重要的属性,再利用支持向量机模型拟合,大大提高了运算效率,使得决策这一类问题变得简便。
[Abstract]:Multi-attribute decision making is also called finite multi-scheme multi-objective decision making, that is, considering multiple schemes, and each scheme is described by multiple attributes. This kind of problem is very common in many fields. Therefore, this research has profound theoretical significance and extensive practical application background. For the multi-attribute decision making problem, we often hope to give the decision maker the best alternative scheme absolutely objectively and fairly. Based on this consideration, this paper puts forward the principle of using support vector machine to fit the decision problem by regression analysis. The mechanism of decision-making model is obtained. Support vector machine (SVM) is a machine learning method based on statistical learning theory. At present, the statistical learning theory is in a stage of popularization to practical application. Support vector machines need to be further improved to meet the needs of practical applications. Support vector machines (SVM), especially for small samples, show good performance for nonlinear problems. In this paper, the support vector machine (SVM) is applied to the multi-attribute decision making problem, and a multi-attribute decision making method based on the support vector machine (SVM) is presented. In this paper, the parameter problem of support vector regression model is studied. In order to select the model parameters more quickly, particle swarm optimization (PSO) algorithm is introduced. The model is applied to a class of multi-attribute decision making problems. The regression fitting of the problem is carried out, and the decision scheme is given. The performance of the optimized model is shown by comparison experiments. Then, the evaluation of journals in multi-attribute decision-making is analyzed experimentally. Because of the strong correlation between the attributes of periodical evaluation problems, some regression fitting methods are restricted, but the support vector regression method can be used to avoid the above problems. Finally, the more complex multi-attribute decision making problem with more attributes is studied. Principal component Analysis (PCA) algorithm is introduced to simplify this kind of problem, and the redundant and unimportant attributes are removed to the maximum extent, and then the support vector machine model is used to fit the problem. The operation efficiency is greatly improved, and the decision making problem becomes simple.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】