Forecasting Indihome Users By Using Trigonometrics, Box Cox, Transformation, Arma Error, Trend, And Seasonal (TBATS) Methods

Abstract

The rapid growth in the use of telecommunication services, particularly IndiHome, has created a pressing need to comprehend and forecast the dynamics of user growth. The swift development of IndiHome services in Pematangsiantar underscores the importance of efficiently forecasting and managing user growth. This research aims to predict future IndiHome users using the TBATS method (Trigonometrics, Box Cox Transformation, Arma Error, Trend, and Seasonal). User data for IndiHome services were obtained from Telekomunikasi Indonesia (Telkom) Pematangsiantar. The collected data underwent analysis and interpolation using the TBATS method. The choice of the TBATS method is attributed to its capability to handle complex patterns of seasonality, trends, and variability in telecommunication service usage data. The TBATS method involves model optimization and model validation using the Mean Absolute Percentage Error (MAPE). The forecasted results for IndiHome users in the next 5 years, with the highest IndiHome users reaching 426,9432, indicate a reasonable forecasting performance. The accuracy of the forecast, as measured by the MAPE calculation, ranges around 30.5%, signifying that the TBATS method demonstrates a satisfactory forecasting capability for IndiHome user data.