呼叫中心到达过程的建模与预测

发布时间：2018-03-28 18:47

本文选题：交互式语音应答到达量　切入点：过离散　出处：《中国科学技术大学学报》2017年09期

【摘要】：呼叫中心到达量的拟合和预测对呼叫中心人力需求计算和排班有着重要作用.根据用户在呼叫中心的不同阶段,可以将到达量分为交互式语音应答(interactive voice response,IVR)到达量和人工服务到达量.国外学者主要对人工服务到达量进行拟合和预测,非齐次泊松过程作为一种模拟到达过程的方法得到广泛使用.而这里首次提出对IVR到达量进行研究,对比实际数据与非齐次泊松过程的均值和方差,发现该呼叫中心IVR到达量呈现"过离散(overdispersion)"现象,不能使用泊松分布拟合.因此选择时间序列模型对IVR到达量进行拟合和预测,用残差的白噪声检验拟合效果,平均绝对误差(mean absolute error,MAE)值判断模型预测优劣.最后使用线性回归模型分析IVR到达量与人工服务到达量之间的关系.结果表明:自回归滑动平均ARIMA(autoregressive integrated moving average)(1,0,1)模型能更好地对该呼叫中心正常天数(normal days,ND)的IVR到达量进行短期预测,而Winters指数平滑法能更好地对春节期间(Spring Festival,SF)的IVR到达量进行短期预测;通过回归模型可以预测呼叫中心人工服务到达量.
[Abstract]:The fitting and prediction of call center arrival is very important to the calculation and scheduling of call center manpower. According to the different stages of call center, The arrival quantity can be divided into interactive voice response voice response (IVR) and manual service arrival. Non-homogeneous Poisson process is widely used as a method to simulate the arrival process. In this paper, the IVR arrival is studied for the first time, and the mean and variance of the actual data and the non-homogeneous Poisson process are compared. It is found that the IVR arrival quantity of the call center is "overdiscrete over dispersion" and can not be fitted by Poisson distribution. Therefore, the time series model is chosen to fit and predict the IVR arrival, and the residual white noise is used to test the fitting effect. The average absolute error and mean absolute error (mae) value are used to judge the model. Finally, the linear regression model is used to analyze the relationship between IVR arrival and manual service arrival. The results show that the autoregressive moving average ARIMA(autoregressive integrated moving average model is better. A short-term prediction of the IVR arrival volume for the normal day of the call center is carried out. The Winters exponent smoothing method can better predict the IVR arrival of Spring FestivalSFs during the Spring Festival, and can predict the arrival of call center artificial services by regression model.
【作者单位】：中国科学技术大学管理学院工商管理系;中国科学技术大学管理学院管理科学系;
【基金】：国家自然科学基金(71501196)
【分类号】：O211.61

【相似文献】