Time Series Forecasting for Incoming Call Volume using LSTM: The case of Ethio Telecom Call Center
No Thumbnail Available
Date
2022-01
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
An accurate forecasting of incoming call volume traffic is key for the operational planning of an inbound call center service. By forecasting, the correct number of incoming calls volumes traffic, can determine staffing and scheduling levels, improve service requirements, and meet customer satisfaction. Currently, in Ethio Telecom inbound type of call center service, the averaging method is used for forecasting incoming call volume traffic. But there is a problem with the averaging method of forecasting the number of incoming calls volume traffic in a call center that cannot handle the trend and seasonality fluctuation. The purpose of this study is to build a model that forecasts incoming call center call traffic using Ethio Telecom call center historical data.
In this thesis, we proposed time series forecasting model for forecasting the incoming call volume traffic. We have used two univariate time series techniques, namely SARIMA and LSTM. Nine months of incoming call center call traffic data is collected from Ethio Telecom. Finally, experimental result indicates that the LSTM model has 24.6% of RMSE improvement of forecasting error compared to the SARIMA model. The overall results of this research work demonstrate that the LSTM model is an effective method for predicting incoming call volume traffic to reflect temporal patterns. Such accuracy is vital to provide a better call center resource allocation for optimization staffing and scheduling problem.
Description
Keywords
call center, time series, Forecasting, Staffing, scheduling, statistical model, deep neural network, LSTM