点击次数：

论文类型：会议论文

发表时间：2017-01-01

收录刊物：CPCI-S

卷号：131

页面范围：394-402

关键字：telephone traffic forecasting; electric system; multi-factors decomposition

摘要：Many factors can affect the telephone traffic of power supply service center. And telephone traffic forecasting would help to estimate the population size, make schedule analysis and copy with the sudden situation. We use some province data of 95598 call center as an instance to analyze factors using time series data as well as extra domain knowledge in order to get the routine rules of telephone traffic. Combining those two to detect the noise and outliers and confirm three factors which respectively are scheduled outage, time, and weekday affecting the traffic flow. The developed time series decomposition model combined with the domain factors is introduced here to make a more accurate relative error, especially when it comes to the rush hour in day time.