Application of Bayesian Optimization to the Temporal Fusion Transformer for Predicting Indonesia’s Carbon Emissions
Journal of Computer Electronic and Telecommunication · 26.10.2026 · 7 (2)
DOI: 10.52435/complete.v7i2.12120 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Computer Electronic and Telecommunication, online erschienen am 26.10.2026. Verfasst von Daffa Ammar Mahendra und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Carbon emissions are one of the main contributors to global climate change, highlighting the need for accurate prediction methods to support decision-making. This study aims to optimize the hyperparameters of the Temporal Fusion Transformer (TFT) using Bayesian Optimization for carbon emission forecasting in Indonesia. The dataset consists of annual carbon emission data from 1970 to 2024, with coal, oil, and gas emissions as input variables and CO₂ emissions as the target variable. The data were normalized using Min-Max Scaling and divided chronologically into training, validation, and testing sets. Bayesian Optimization was conducted using 20 trials for each random seed using a search space covering learning rate, hidden size, dropout rate, batch size, and epoch, with a sequence length of four time steps. Performance was evaluated using MAE, MAPE, and SMAPE across four random seeds. The results show that Bayesian Optimization reduced the average MAE from 52.3874 ± 2.4281 MtCO₂ to 46.7754 ± 5.5239 MtCO₂, while MAPE decreased from 7.13% ± 0.35% to 6.40% ± 0.83% and SMAPE from 7.47% ± 0.41% to 6.69% ± 0.82%. The model also captured the general trend of Indonesia’s carbon emissions during the 2020-2024 testing period, indicating improved prediction performance.
Zitieren
APA
Mahendra, D. A., Muttaqin, F., & Via, Y. V. (2026). Application of Bayesian Optimization to the Temporal Fusion Transformer for Predicting Indonesia’s Carbon Emissions. Journal of Computer Electronic and Telecommunication, 7(2). https://doi.org/10.52435/complete.v7i2.12120
Vancouver
Mahendra DA, Muttaqin F, Via YV. Application of Bayesian Optimization to the Temporal Fusion Transformer for Predicting Indonesia’s Carbon Emissions. Journal of Computer Electronic and Telecommunication. 2026;7(2). doi:10.52435/complete.v7i2.12120
BibTeX
@article{Mahendra2026,
author = {Mahendra, Daffa Ammar and Muttaqin, Faisal and Via, Yisti Vita},
title = {Application of Bayesian Optimization to the Temporal Fusion Transformer for Predicting Indonesia’s Carbon Emissions},
journal = {Journal of Computer Electronic and Telecommunication},
year = {2026},
volume = {7},
number = {2},
doi = {10.52435/complete.v7i2.12120}
}
EUVIMED