研究商业银行客户分类优化问题.商业银行客户类别具有多变性,其类别由初始聚类中心来确定,而传统K均值初始聚类中心固定,不能适应客户类别具有多变性,导致商业银行客户分类结果易陷入局部最优,分类准确率极低.为了提高商业银行客户分类的准确率,提出粒子群优化K均值聚类的商业银行客户分类模型.模型将K均值的初始聚类中心作为一个粒子,商业银行客户分类准确率作为粒子群优化的目标函数,通过粒子相互协作获得最优初始聚类中心,聚类中心具有自适应性,使然后采用最优K均值聚类算法对银行客户进行分类.仿真结果表明,优化K均值算法收敛速度快,提高了客户分类准确率,分类结果更加合理,便于对商业银行为客户采取相应经营策略.%Study on bank customer subdivision. In the K - means algorithm, random choosing of initial values may cause different clustering results, and sometimes it may even have no results. Additionally, this algorithm is based on gradient descent, therefore, it may inevitably get into the local extreme optimization frequently. After analy- zing the disadvantages of the classical K - means clustering algorithm, an improved K - means is proposed based on Particle Swarm Optimization algorithm. The theory analysis and experimental results show that the algorithm has greater searching capability, and avoids the local optima. The application of this improved K - means algorithm of subdivision of bank customer can improve the bank customer classification accuracy, the clustering results are more reasonable, and it can provide the reasonable decision support for bank managers.
展开▼
