摘要
树形算法由于其对大量高维数据的有效处理、对噪声点的高容忍度和对知识的有效表示,是最常用的CRM客户细分技术。通过对几类树形算法,包括决策树C4.5算法、决策树CART算法和平衡随机森林BRF算法,在解决电信客户细分问题中的表现进行分析研究,并且选用BP神经网络算法作为树形算法的参照,最终研究得出:平衡随机森林在处理电信客户问题上具有最好的表现。
Due to the effective processing of large amounts of high-dimensional data, high tolerance for noise and effective representation of knowledge, tree algorithm is the most common CRM customer segmentation technique. The performance of tree algorithm, including the C4.5, the CART and the balanced random forest, in solving telecommunication customer segmentation problems is analyzed. BP neural network algorithm is compared. Experiments have shown that balanced random forest has the best performance in dealing with the problem.
出处
《计算机时代》
2014年第5期1-4,共4页
Computer Era
