基于支持向量机的模糊回归估计

针对给定的大规模数据集的回归估计问题，提出基于支持向量机的模糊回归估计方法.该方法把复杂的数据集看作多个群体的混合，每个群体采用单一的回归模型进行描述，使得大规模数据集的回归估计问题变成了一个多模型估计问题.在此基础上把支持向量机与模糊C聚类结合起来得到基于支持向量机的模糊回归模型，并给出了实现该模型回归估计的算法.该方法对大规模的数据样本进行模糊C聚类，并回归估计各聚类的数据样本.数值仿真结果表明，该方法在聚类数据样本的同时能实现多个模型的回归估计，而且模糊隶属度的初始化影响要小于其他的模糊回归估计方法.

Fuzzy regression estimation based on support vector machine

To solve the problem of regression estimation for large dataset, a fuzzy regression estimation method based on support vector machine (SVM) was proposed. The large dataset was viewed as a mixture of multiple populations, and each population was represented by a single regression model. The problem of regression estimation for large dataset was viewed as a problem of multiple regression model estimation. Fuzzy regression estimation based on SVM was used to estimate the multiple regression models, and the algorithm for solving fuzzy regression model based on SVM was provided. A numerical method was used to fit nonlinear models for mixed dataset. Simulation result shows that the method of fuzzy regression estimation based on SVM can discriminate the multiple regression models with a fuzzy partition of dataset while fitting perfectly these models. With this method, the influence caused by the initialization of fuzzy membership is less than that of other fuzzy regression estimation methods.