一种基于时间序列算法的资金流入流出预测模型  

A prediction model of funds inflow and outflow based on time series algorithm

作　　者：曹璨[1] 黄海[1] 徐可[1] 

机构地区：[1]中国科学技术大学信息科学技术学院,安徽合肥230027

出　　处：《微型机与应用》2017年第14期52-56,共5页Microcomputer & Its Applications

摘　　要：资金的流入流出预测对于降低网络金融平台的流动性风险、提高资金利用率具有重要意义。根据资金流入流出历史数据,对蚂蚁金服公司余额宝资金未来30天流入流出的预测问题进行研究。由于历史数据不稳定且多噪声,首先采用序列转换方法对不平稳序列进行差分处理提高其数据稳定性,其次对该序列进行模型识别和参数估计,利用时间序列模型初步预测,并对残差序列进行模型检验,最后利用通过检验的模型预测结果。实验结果表明,此模型可以有效地对余额宝用户的资金流入流出金额进行预测。It is great significant to reduce the liquidity risk of the network financial platform and improve the funds utilization rate to predict funds inflow and outflow. In this paper, we research on predicting Ant Financial Services Group YuEBao funds inflow and outflow in next 30 days according to historical records. Because historical data is instable with much noise, we utilize the differential conversion method to transfer the original sequences to stable sequence. Then, though the model identification and parameter estimation we obtain a preliminary model. Next we select the suitable time series model to predict the residual sequence and do model check. At last, we use the approved model trains data to predict final results. The results of experiments verify that the proposed prediction model can effectively predict the funds inflow and outflow.

关 键 词：资金流入流出预测 序列转换 参数估计 时间序列模型 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]