Implementation of Long Short Term Memory Model in Forecasting Internet Service Sales

Abstract

Competition in providing internet services in Indonesia is getting tougher. Market demand that is increasingly complicated to predict makes companies have to work more to satisfy customers. The application of forecasting methods for client needs can be a solution. Machine Learning-based forecasting with the Long Short Term Memory (LSTM) method can be one way of making forecasts. The output of this research is the forecasting of the price of the service product which is expected to make the company take policies to take actions that can minimize losses for the client and the company. In this study, the author will use the Long Short Term Memory (LSTM) method to predict the price of internet services at the Hypernet Indodata company using time series data. The data used is internet service sales in 2016–2018 obtained from PT. Hypernet Indodata. The results obtained in this study resulted in a Root Mean Square Error (RMSE) value of 8.7463 and a Mean Absolute Percentage Error (MAPE) of 4.167% indicating that the LSTM model already has the right configuration and is successful in predicting service prices quite well.