COMPARISON OF VISITS TO THE FACEBOOK WEBSITE USING THE FACEBOOK PROPHET AND SARIMA MODELS

Abstract

Facebook is widely used across various segments of society and offers significant potential as a marketing tool due to its diverse content and communication features. This study aims to forecast visits to the Facebook website for the period of March 2023 to February 2024 using two forecasting methods: the Facebook Prophet and SARIMA models. We analyzed 167 data points representing the percentage of visits to the Facebook website from April 2009 to February 2023, sourced from Statcounter. Our forecasting results indicate that the Facebook Prophet model yielded a Mean Absolute Percentage Error (MAPE) of 22.3% and a Root Mean Square Error (RMSE) of 16.2%. In comparison, the SARIMA model achieved a MAPE of 6.31% and an RMSE of 5.02%. Based on these accuracy metrics, the SARIMA model is determined to be more suitable for forecasting the percentage of visits to the Facebook social media website in Indonesia. This finding underscores the importance of selecting appropriate forecasting models for accurate predictions in social media analytics