组合分类器在金融行业数据分类中的应用研究

摘要

金融行业每天都会产生大量的数据,如何有效利用这些数据是每个金融企业都应该考虑的问题,但是目前存在的分类方法或多或少都存在一定的缺陷.对此,在C5.0、logistic和贝叶斯三种分类方法的基础上提出一种基于置信度加权的组合分类模型,并与三种分类算法进行比较分析.结果表明,组合分类器模型的分类表现最好.对于组合分类器的运用,可有效提高分类准确率,规避单一分类器的分类缺陷.%The financial industry produces huge data every day,each financial enterprise should consider how to use these data effectively,but the current classification method are still flawed.Thus,a composite classification model based on the confidence weighting of C5.0,logistic and Bayesian classification methods is proposed and compared with other three classification method.The results show that the combined classifier model performed best.The combined classifier effectively improve the classification accuracy and avoid the classification defects of single classifier.