基于最小二乘支持向量机的数据挖掘应用研究

Data Mining Based on Least Squares Support Vector Machines

随着数据仓库技术、联机分析技术的发展,基于数据库的数据挖掘已成为一种重要的数据处理手段。最小二乘支持向量机作为一种新的机器学习方法,具有全局收敛性和良好的泛化能力,本文将其应用于数据挖掘的分类与预测研究,通过核函数的选择及参数优化,并结合支持向量机、多层感知器神经网络模型及判别分析方法进行比较研究,证明最小二乘支持向量机作为一种有效的数据挖掘算法具有较高精度。

With the development of data warehouse and OLAP technology, data mining is becoming an impartment measure of data processing. Least Squares Support Vector Machines （IS- SVM）, as a new machine learning method, has such properties as global convergence and good ability of extension. The research applies IS- SVM to data classification and prediction in terms of data mining. By kernel function selecting and parameter optimizing, the performance has been evaluated and compared with SVM, MLP neural network and discriminant analysis.