APPLYING MACHINE LEARNING APPROACHES FOR NETWORK TRAFFIC FORECASTING

Abstract

In the era of the digital world. The communication and the use the internet is an important role in today's society. As a result, the number of users networks traffic increases but not enough resources for users causing users to receive inefficient services. Therefore, the network service provider must take action to fix the aforementioned problem. Forecasting is therefore necessary in order to determine the amount of network traffic in order to support the future increase in user numbers. Consequently, this research investigates to assess network traffic forecasts comparing the machine learning: Artificial Neural Network (ANN), Long Short-Term Memory (LSTM), and statistical methods: Autoregressive Integrated Moving Average (ARIMA), Simple Moving Average (SMA). The method of sliding window will be used simultaneously and evaluate the forecast and model performance using the MAE, MAPE, MSE, RMSE and R-square algorithms, respectively. The results show that machine learning forecasting is more effective than statistical forecasting. Because the error value is lower, the model can reliably anticipate data. Therefore, the results of this research are expected to help network service provider to improve their networks quickly and efficiently to accommodate the number of users that may increase in the future.

Topics & Concepts