Jibran, Kemal Fahreza (2026) PREDICTION OF AIR TEMPERATURE IN SURABAYA CITY USING PROPHET WITH GRID SEARCH HYPERPARAMETER OPTIMIZATION. Undergraduate thesis, Universitas Pembangunan Nasional "Veteran" Jawa Timur.
|
Text (Cover)
Cover fix.pdf Download (791kB) | Preview |
|
|
Text (Bab 1)
Chapter 1 Repository.pdf Download (239kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Bab 2)
Chapter 2 Repository.pdf Restricted to Repository staff only until 3 June 2029. Download (494kB) |
|
|
Text (Bab 3)
Chapter 3 Repository.pdf Restricted to Repository staff only until 3 June 2029. Download (904kB) |
|
|
Text (Bab 4)
Chapter 4 Repository.pdf Restricted to Repository staff only until 3 June 2029. Download (1MB) |
|
|
Text (Bab 5)
Chapter 5 Repository.pdf Download (226kB) | Preview |
|
|
Text (Daftar Pustaka)
References Repository.pdf Download (264kB) | Preview |
|
|
Text (Lampiran)
Appendices Repository.pdf Restricted to Repository staff only Download (2MB) |
Abstract
Global climate change has led to rising air temperatures with significant impacts on urban areas such as Surabaya, particularly due to the Urban Heat Island phenomenon. These conditions necessitate accurate temperature prediction methods to support informed decision-making. This study aims to analyze the performance of the Prophet model and to measure the effect of hyperparameter optimization using the Grid Search method on the accuracy of daily air temperature prediction in Surabaya. The data utilized consist of daily average temperatures covering the period from January 1, 2020 to December 31, 2025, obtained from timeanddate.com. This study employs a time series approach within the OSEMN framework, encompassing data acquisition, data cleaning, exploration, modeling, and interpretation. The Prophet model was first constructed as a baseline, and subsequently optimized using Grid Search across the parameters changepoint_prior_scale, seasonality_prior_scale, and seasonality_mode. Model performance was evaluated using the RMSE, MAE, and MAPE metrics. The default Prophet baseline model yielded an RMSE of 0.8684°C, an MAE of 0.6602°C, and a MAPE of 2.3259%. Following Grid Search optimization with the best-performing configuration of changepoint_prior_scale = 0.1, seasonality_prior_scale = 20, and seasonality_mode = additive, the optimized model produced an RMSE of 0.8584°C, an MAE of 0.6580°C, and a MAPE of 2.3114%. The results demonstrate that the optimized model outperforms the default model, as evidenced by a reduction in error values across all evaluation metrics. The best-performing model proved more capable of capturing temperature trend and seasonal patterns, thereby producing predictions that are more stable and adaptive to the data. It is therefore concluded that hyperparameter optimization via Grid Search consistently improves the predictive performance of the Prophet model across all evaluation metrics, although the magnitude of improvement is incremental, given the already stable performance of the Prophet baseline model.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics Q Science > QA Mathematics > QA76.6 Computer Programming |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Kemal Fahreza | ||||||||||||
| Date Deposited: | 03 Jun 2026 08:23 | ||||||||||||
| Last Modified: | 03 Jun 2026 08:23 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/53612 |
Actions (login required)
 |
View Item |