A Comparison Of The Bayesian Structural Time Series Technique With The Autoregressive Integrated Moving Average Model For Forecasting

Abstract

In this work, two forecasting models, Bayesian Structural Time Series (BSTS) and Autoregressive Integrated Moving Average (ARIMA), were used to estimate Turkish coal production from 1970 to 2022 which obtained through the World Bank database. The main objective was to evaluate these models' predictive accuracy for trends in coal production. The modeling and analytic procedures were carried out using the R program, and MAE, RMSE, R² and MAPE were also used for this comparison in order to guarantee accurate and thorough results. The results showed that when it came to estimating the time series data of coal production in Turkey, the BSTS model outperformed the ARIMA model. When compared to the Box-Jenkins approach of the ARIMA model, the BSTS model's which takes into account the inherent uncertainties and complexities in coal production showed superior accuracy and reliability.