Performance Comparison Between Facebook Prophet and SARIMA on Indonesian Stock

This paper focuses on comparing the performance of two supervised learning methods, SARIMA and Facebook Prophet, in predicting stock prices of Indonesian companies. SARIMA, an extension of the ARIMA model, incorporates seasonal components in the data meanwhile Facebook Prophet is a business forecasting model that incorporates trend, seasonality, and holiday components in its approach. The study analyzes five sets of stock data and measures accuracy using the R-square method. For SARIMA, the data undergoes stationarity testing, model selection with hyperparameter tuning, and prediction of stock prices. Facebook Prophet involves dividing the data into training and testing sets, fitting the model, and making predictions using various components. The comparison is based on the R-square values of the predicted data. The findings reveal that the SARIMA method outperforms Facebook Prophet in predicting stock prices for the selected Indonesian companies, as indicated by higher R-square values obtained from SARIMA.