EUVIMEDEUVIMED entdecken
Open Access · CC BY-SA 4.0

Air Temperature Prediction of Surabaya City Using Prophet with Random Search Hyperparameter

Kemal Fahreza Jibran, Rizky Parlika, Wahyu Syaifullah Jauharis Saputra

bit-Tech · 10.08.2026 · 9 (1) · 186-197

DOI: 10.32877/bt.v9i1.3824 ↗

Kurz gefasst

Fachartikel in der Zeitschrift bit-Tech, online erschienen am 10.08.2026. Verfasst von Kemal Fahreza Jibran und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Air temperature prediction plays a crucial role in supporting urban sustainability, public health, and climate adaptation, particularly in tropical cities experiencing rapid urbanization such as Surabaya. Despite the growing adoption of time-series forecasting models, the effectiveness of systematic hyperparameter optimization for improving Prophet-based temperature prediction in tropical urban environments remains insufficiently explored. This study aims to investigate whether Random Search hyperparameter optimization can significantly enhance the forecasting performance of the Prophet model for urban air temperature prediction. A time-series dataset consisting of 2,182 daily average temperature observations collected from the Surabaya/Perak Meteorological Station between January 2020 and December 2025 was utilized. The proposed methodology includes data preprocessing, chronological data splitting (70:15:15), model training, and evaluation using RMSE, MAE, and MAPE. Random Search was applied to optimize two key hyperparameters, namely changepoint_prior_scale and seasonality_prior_scale, over 20 iterations. The optimized model achieved RMSE, MAE, and MAPE values of 0.8937°C, 0.7076°C, and 2.46%, respectively, significantly outperforming the baseline Prophet model, which recorded 2.2507°C, 2.0640°C, and 7.28%. These findings demonstrate that even a computationally efficient optimization strategy such as Random Search can substantially improve Prophet’s predictive accuracy. Compared to commonly reported results in temperature forecasting studies, the proposed approach achieves competitive performance while maintaining model interpretability. This study contributes by providing an effective and practical framework for enhancing time-series forecasting in tropical urban environments using optimized Prophet models.

Zitieren

APA

Jibran, K. F., Parlika, R., & Saputra, W. S. J. (2026). Air Temperature Prediction of Surabaya City Using Prophet with Random Search Hyperparameter. bit-Tech, 9(1), 186-197. https://doi.org/10.32877/bt.v9i1.3824

Vancouver

Jibran KF, Parlika R, Saputra WSJ. Air Temperature Prediction of Surabaya City Using Prophet with Random Search Hyperparameter. bit-Tech. 2026;9(1):186-197. doi:10.32877/bt.v9i1.3824

BibTeX
@article{Jibran2026,
  author = {Jibran, Kemal Fahreza and Parlika, Rizky and Saputra, Wahyu Syaifullah Jauharis},
  title = {Air Temperature Prediction of Surabaya City Using Prophet with Random Search Hyperparameter},
  journal = {bit-Tech},
  year = {2026},
  volume = {9},
  number = {1},
  pages = {186--197},
  doi = {10.32877/bt.v9i1.3824}
}
Kostenlos starten →