Comparison of Moving Average and Exponential Smoothing Methods for Estimating the Number of Internet Service Disruptions

Abstract

Fairly tight competition in telecommunication business requires companies to continue improving the quality of their services, including handling network problems or customer complaints. However, the amount of disturbance that occurs is often unpredictable. Indeed, this condition will impact the allocation of resources and budget for handling customer disturbances in the following period, with uncertainty. This research aims to apply the moving average (MA) and single exponential smoothing (SES) forecasting methods to predict the number of internet service disruptions. The case study used in this paper is an internet service provider company with service coverage in the northern part of Surabaya. The number of disturbances data was collected from January 2022 to May 2022, with 18,453 data in total. The disturbances can be divided into five types: physical, mass, logical, PSB/migration, and others. Forecasting is carried out using the MA method with a period of three months. Meanwhile, forecasting using the SES method was carried out by first determining the alpha value that produces the smallest Mean Absolute Percentage Error (MAPE) value. The analysis results show that both forecasting methods are relatively effective and efficient in estimating the number of disturbances. Forecasting performance testing was carried out by measuring the Mean Absolute Deviation (MAD), Mean Square Error (MSE), and Mean Absolute Percentage Error (MAPE) values. The results of forecasting performance measurements show that the SES method is much better than MA for all types of disturbance data, with MAPE values ​​below 2%.