非平衡数据集的支持向量域分类预测模型研究

Research on Support Vector Domain Classification Predication Model of Non-Balance Data Set

基于非平衡数据集的支持向量域分类模型,提出了一种银行客户个人信用预测方法。首先分析了信用预测的主要方法及其不足,然后研究了支持向量域分类模型及其参数的非负二次规划乘性更新算法,进而提出基于支持向量域分类模型的银行客户个人信用预测方法,最后使用人工数据和实际数据对提出方法与支持向量机预测方法进行对比实验。实验结果表明对于银行客户个人信用预测的非平衡数据分析问题,基于支持向量域模型的分类预测方法更有效。

A new predication method of customer credit of banks is proposed based on the support vector domain classification model of non-balance data set. Main predication techniques of customer credit and their shortcoming are reviewed firstly. Following the support vector domain classification model is analyzed. And the muhiplicative updating principles of the parameters using nonnegative quadratic programming are investigated. The predication method of customer credit of banks based on the support vector domain model is proposed further. At last, we compare experiments on synthesized and real data using proposed method and the support vector machine predication method. The experimental results show that the proposed method is more effective than the support vector machine predication method for the problem of non-balance data classification such as predication of customer credit of banks.