基于MARS的肝阳化风证的模型研究

发布时间：2018-04-10 22:00

本文选题：高维数据 + 高维建模　；参考：《湖南师范大学》2015年硕士论文

【摘要】：多元自适应回归样条是一种基于拟合且自适应性很强的方法,同时也是一种非常灵活的、泛化能力很强的、非线性的、非参数的专门针对高维数据的处理方法。该方法以样条函数的张量积作为基函数,以样本数据坐标作为可选的节点矢量值,算法过程以广义交叉验证最小为标准选择基函数和节点矢量。本文紧紧围绕高维建模展开研究,并试图探讨所建立的高维模型。中医肝阳化风证是肝风内动证的三个亚型之一。为了探讨本证的病理生理学基础,前人运用文献研究积累、专家经验、数理统计分析等多种方法做了大量的研究,但是其本质内涵仍不清晰。肝阳化风证具备中医的复杂性、整体性、非线性、动态性等各种特点,决定了肝阳化风证的研究也需要走复杂系统研究的道路。肝阳化风证的辩证和诊疗标准较多,对其进行量化后呈现出高维的状态,欲从科学理论的角度揭示其辩证规律需要用到现代建模理论的思想和方法。基于上述思想的指导,本文对肝阳化风证进行分析,利用多元自适应回归样条算法建立最佳预测模型。本文主要研究内容如下：(1)分析了多元自适应回归样条方法区别于传统方法在高维建模中的优势,其良好的模型解释能力,明确的指出了变量的分类,模型的解释在方差分解下更加直观；(2)通过对症状数据的量化及规范化,根据中医系统的复杂性,挖掘变量之间的关系,寻找隐含的证型模型,然后分析中医问诊的症状和证型、症状和症状之间内在结构关系来建立模型；(3)运用模型的方差分解图形直观的观察预测变量,分析变量对模型的影响方式预测症状对肝阳化风证的影响方式；(4)利用预留的数据检验模型,依据指定的各项指标来评价模型的精度。结果显示建立的肝阳化风证模型有着较好的预测精度。
[Abstract]:Multivariate adaptive regression spline is a method based on fitting and strong adaptability, and it is also a very flexible, generalizable, nonlinear, non-parametric processing method for high-dimensional data.In this method, the tensor product of the spline function is taken as the basis function, the coordinate of the sample data is taken as the optional node vector value, and the basis function and the node vector are selected by using the generalized cross-validation minimum as the standard.This paper focuses on the research of high-dimensional modeling and attempts to explore the established high-dimensional model.Traditional Chinese medicine liver-yang syndrome is one of the three subtypes of the liver wind internal movement syndrome.In order to explore the pathophysiological basis of this syndrome, many methods such as literature research, expert experience, mathematical statistics and so on have been used to do a lot of research, but its essential connotation is still unclear.Because of the complexity, integrity, nonlinearity, dynamics and other characteristics of traditional Chinese medicine, the study of liver-yang wind syndrome also needs to take the road of complex and systematic research.There are many dialectical and diagnostic criteria for liver-yang and wind syndrome, which are characterized by high dimension after quantification. To reveal its dialectical law from the angle of scientific theory, it is necessary to use the thought and method of modern modeling theory.Based on the above thought, this paper analyzes the syndrome of liver yang and wind, and sets up the best prediction model by using multivariate adaptive regression spline algorithm.The main contents of this paper are as follows: (1) this paper analyzes the advantages of the multivariate adaptive regression spline method compared with the traditional method in high-dimensional modeling, its good model interpretation ability, and clearly points out the classification of variables.By quantifying and standardizing the symptom data, according to the complexity of TCM system, excavating the relationship between variables, looking for the implicit syndrome model, and then analyzing the symptoms and syndrome types of TCM inquiry, the explanation of the model is more intuitionistic under the decomposition of variance.The internal structural relationship between symptoms and symptoms is used to establish model No. 3) using the variance decomposition pattern of the model to observe the prediction variables intuitively.Analysis of the influence of variables on the model predicting the influence of symptoms on the syndrome of liver-yang and wind. (4) using the reserved data test model to evaluate the accuracy of the model according to the specified indicators.The results showed that the model of liver Yang and wind syndrome had good prediction accuracy.
【学位授予单位】：湖南师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：O212.1

【相似文献】