摘要
为得到准确的前瞻性话务量数据,有效解决呼叫中心资源合理分配和针对商旅类呼叫中心话务量的预测问题,建立了卡尔曼滤波器预测模型。依据卡尔曼滤波器算法特点,以及话务量数据序列周期性和季节性特性的分析结论,确定了卡尔曼滤波器日话务量预测模型的状态转移矩阵和观测矩阵。提出了以周为基本时间周期提取训练数据和预测话务量,根据历史话务量,通过预定义的高峰期话务量差异数据库修正预测结果的算法思想。MATLAB预测试验结果显示日话务量预测结果比较好的逼近实际日话务量。
To accurately predict the telephone traffic and help to manage the call center resources effectively and reasonably, a telephone traffic prediction algorithm based on the Kalman filtering theory is proposed in this paper. The state transfer matrix and observation matrix of Kalman filter is defined according to the Kalman filtering theory and the fact of conclusion that the telephone traffic is periodic in weeks and seasons. In this paper, the telephone traffic data for training the model is collected by the day of week, and the prediction model will give cyclical telephone traffic prediction in cycle of week. According to the historical traffic data saved in database, the prediction output of the model will be tuned by a predefined offset when it is in hot-time with peak traffic. The MATLAB experiments show that the prediction result of a day fits well with the actual daily traffic data.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第12期4405-4409,共5页
Computer Engineering and Design
关键词
卡尔曼滤波
呼叫中心
话务量预测
季节
周期
Kalman filter
call center
telephone traffic forecast
seasonal
cyclical
