Peramalan Jumlah Kunjungan Wisatawan Mancanegara Ke Jawa Timur Menggunakan Metode Hybrid ARIMA Intervensi Multi Input dan Long Short-Term Memory (LSTM)

Nasaruddin, Aqilah Mutiara (2026) Peramalan Jumlah Kunjungan Wisatawan Mancanegara Ke Jawa Timur Menggunakan Metode Hybrid ARIMA Intervensi Multi Input dan Long Short-Term Memory (LSTM). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221103-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Penelitian ini bertujuan untuk menganalisis karakteristik data, merumuskan model terbaik, serta melakukan peramalan jumlah kunjungan wisatawan mancanegara ke Jawa Timur menggunakan pendekatan Hybrid ARIMA intervensi multi-input dan LSTM. Data yang dianalisis dalam penelitian ini berupa data deret waktu bulanan jumlah kunjungan wisatawan mancanegara ke Jawa Timur, dengan rentang periode Januari 2015 hingga Desember 2024. Sebelum terjadinya pandemi COVID-19, data menunjukkan pola fluktuatif, sebelum akhirnya terjadi penurunan ekstrem akibat intervensi COVID-19 pada tahun 2020, kemudian menunjukkan tren pemulihan secara bertahap hingga 2024. Kondisi tersebut mengindikasikan terjadinya efek intervensi yang signifikan, sehingga dibutuhkan metode yang mampu mengakomodasi perubahan tersebut sekaligus menangkap karakteristik non linier yang tersisa pada residual. Pemodelan dilakukan dengan mengombinasikan metode ARIMA intervensi multi-input untuk menangkap pola linier dan efek intervensi dan Long Short-Term Memory (LSTM) untuk menangkap pola non linier pada residual. Model hybrid terbaik diperoleh dari kombinasi ARIMA([4],1,0) dengan orde intervensi pertama (2,0,1) dan intervensi kedua (0,0,[8]), serta LSTM dengan hyperparameter lag input sebesar 13, 16 unit, 1 layer, dropout 0,35, learning rate 0,05, dan batch size 8. Model ini mampu menangkap pola linier, efek intervensi, dan pola nonlinier residual, sehingga menghasilkan nilai MAPE akhir sebesar 20,78%. Selanjutnya, model hybrid yang telah terbentuk diaplikasikan untuk meramalkan jumlah kunjungan wisatawan mancanegara ke Jawa Timur periode Januari hingga Juni 2025. Hasil peramalan menunjukkan bahwa model ini mampu mengikuti pola historis serta menangkap fluktuasi kunjungan wisatawan.
======================================================================================================================================
This study aims to analyze data characteristics, formulate the best model, and forecast the number of international tourist visits to East Java using a Hybrid ARIMA multi-input intervention and LSTM approach. The data analyzed in this study consists of monthly time series data on the number of international tourist visits to East Java, covering the period from January 2015 to December 2024. Prior to the COVID-19 pandemic, the data showed a fluctuating pattern, followed by a sharp decline due to COVID-19 interventions in 2020, and then a gradual recovery trend through 2024. These conditions indicate a significant impact from the interventions, necessitating a method capable of accommodating these changes while capturing the remaining nonlinear characteristics in the residuals. Modeling was performed by combining a multi-input ARIMA intervention model to capture linear patterns and intervention effects with a Long Short-Term Memory (LSTM) model to capture nonlinear patterns in the residuals. The best hybrid model was obtained from a combination of ARIMA([4],1,0) with first-order intervention (2,0,1) and second-order intervention (0,0,[8]), as well as LSTM with input lag hyperparameters of 13 and 16 units, 1 layer, dropout of 0.35, a learning rate of 0.05, and a batch size of 8. This model is capable of capturing linear patterns, intervention effects, and nonlinear residual patterns, resulting in a final MAPE value of 20,78%. Subsequently, the developed hybrid model was applied to forecast the number of international tourist visits to East Java for the period from January to June 2025. The forecasting results indicate that this model is capable of following historical patterns and capturing fluctuations in tourist visits.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Forecasting, International Tourist, LSTM, Multi-Input Intervention, ARIMA, Intervensi Multi-input, Peramalan, Wisatawan Mancanegara
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Aqilah Mutiara Nasaruddin
Date Deposited:	30 Jul 2026 03:58
Last Modified:	30 Jul 2026 03:58
URI:	http://repository.its.ac.id/id/eprint/140377

Actions (login required)

View Item