PREDICTION OF THE NUMBER OF TOURIST VISITS IN EAST JAVA PROVINCE USING THE GATED RECURRENT UNIT (GRU) METHOD

Karim, Ahsan Nashihul (2026) PREDICTION OF THE NUMBER OF TOURIST VISITS IN EAST JAVA PROVINCE USING THE GATED RECURRENT UNIT (GRU) METHOD. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (cover) 20081010106.-cover (1).pdf Download (878kB)
	Text (bab 1) 20081010106.-bab1.pdf Download (147kB)
	Text (bab 2) 20081010106.-bab2.pdf Restricted to Repository staff only until 18 July 2028. Download (358kB) \| Request a copy
	Text (bab 3) 20081010106.-bab3.pdf Restricted to Repository staff only until 18 July 2028. Download (410kB) \| Request a copy
	Text (bab 4) 20081010106.-bab4.pdf Restricted to Repository staff only until 18 July 2028. Download (1MB) \| Request a copy
	Text (daftar pustaka) 20081010106.-daftarpustaka.pdf Download (140kB)
	Text (bab 5) 20081010106.-bab5.pdf Download (124kB)

Abstract

The tourism sector in East Java Province faces serious challenges in predicting tourist visit numbers due to fluctuations driven by holiday seasons, religious celebrations, and external shocks such as the COVID-19 pandemic. Conventional methods previously used by the East Java Provincial Tourism Office, namely trend extrapolation and simple historical averages, have proven inadequate for capturing seasonal patterns or producing accurate projections at the regency/city level. This study applies the Gated Recurrent Unit (GRU) method to build a prediction model for the number of domestic tourist visits per regency/city in East Java Province, using monthly data from the Central Bureau of Statistics (BPS) for the period January 2019 to December 2025, covering all 38 regencies/cities with a total of 3,192 data points. The model was developed using an integrated panel data approach with a sliding window technique (window size of 12 months), Min-Max Scaler normalization, and a time series split into training data (2019–2024) and testing data (2025). The model architecture consists of a GRU layer with 64 units, a dropout layer with a rate of 0.8, and a dense layer. Training showed good convergence with a train loss of 0.015233 and a best validation loss of 0.014702, with no indication of overfitting. Based on hyperparameter testing scenarios, the best configuration (S4-C) achieved an RMSE of 95,216.35 using GRU units of 64, window size of 12 months, learning rate of 0.001, dropout rate of 0.8, 100 epochs, and batch size of 32. Per-region evaluation showed the model performed best in cities with stable visitation patterns such as Mojokerto City (RMSE 18,434) and Probolinggo City (RMSE 23,410), while regions with extreme fluctuations such as Malang Regency and Surabaya City produced higher RMSE values. The results indicate that the GRU method is effective for tourist visit prediction in East Java Province and has strong potential as a data-driven strategic decision-making tool for regional tourism management.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Al Haromainy, Muhammad Muharrom	NIDN0701069503	muharrom.if@upnjatim.ac.id
Thesis advisor	Rakhmadi, Ardhon	199108052024061002	Ardhon.if@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA76.87 Neural computers

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Ahsan Nashihul Karim

Date Deposited:

22 Jul 2026 04:00

Last Modified:

22 Jul 2026 04:00

URI:

https://repository.upnjatim.ac.id/id/eprint/55416