摘要
基于多分类器融合技术,建立了新的客户分类模型,该模型通过使用分类融合器,将多个分类器得到的客户信用评估结果进行合并,从而综合了不同分类器的局部优势,提高了分类性能。采用线性分类融合器,并通过遗传算法对分类器进行优化。实验分析表明,该方法的分类效果明显优于传统的运用单个分类器的分类方法。
By combining multiple classifiers, a new model for customer segmentation was established in this study. The classification ability was improved by combining multiple results of different classifiers into a single new result. Genetic algorithm was adopted to optimize the linear fusion model. Experimentation results showed that the classification ability of this new method is better than tradition customer segmentation methods.
出处
《管理科学》
CSSCI
2004年第2期64-67,共4页
Journal of Management Science
基金
国家自然科学基金资助项目(7017013)
黑龙江省自然科学基金资助项目(G0304)
关键词
多分类器融合
遗传算法
人工神经网络
客户细分
Multiple classifiers syncretizing
Genetic algorithm
Artificial neural networks
Customer segmentation
