银行存款模型及应用分析

近年来,越来越多的国际知名企业都认识到了从原始数据中寻找规律对决策管理的重要性.一些顶尖级的银行产品企业,象IBM,Oracle,Informix和Sybase等,已经开发了许多用于银行数据挖掘的软件产品,并为国际上的一些著明银行建立了高精度的统计模型以支持银行管理.存款是银行评价业绩的一项重要指标.建立高精度的存款模型有利于银行的日常资金管理,能提高银行的资金利用率,降低成本等.本文以国内某大城市两大银行的实际业务数据为背景,给出了银行存款模型的建立过程,并分析了模型的应用.本文的一些有趣结果对时间序列建模有一定的启示.

DEPOSIT AMOUNT MODELSAND FORECASTING ANALYSIS FOR BANKS

In recent years,more and more world famous enterprises began to realize the importance of finding relationships from the original data. The first rank computer company like IBM,Oracle,Informix and Sybase have already made a progress in this kind of data mining software techniques,They designed highly precision statistical models for some world famous banks,department company etc[5]. The remaining sums of deposits are one of the most important indexes to evaluate the achievements of a bank. It will be very heplful to its management of business if a bank can predict the coming remaining sums of current deposits accurately. In this paper, we build three types of prediction model for the remaining sums of current deposits of two banks[6]. Comparing the models we provide in this paper, we get some interesting conclusions, which may be useful experiences for reference when we do time series studies in the future. The conclusions include:●Though very different in forms ,the prediction models can all get high precision if we fit each model appropriately.●Each model has its own specific requirements of data set and independent variables.●The best-selected forms of the three prediction maodels remain the same despite of the changes of prediction periods.●The most suitable model for long-term prediction is an AR-GARCH model.The original data sets used in this paper are pre-checked to find that if there are repeated data registered(if yes ,then only keep the first),or very large or small data caused by recording errors.The Software used herein is SAS.