Prediksi Lonjakan Penumpang Kereta Api Menggunakan Bayesian Extreme Value Theory

Putri, Renata Najiah Efandy (2026) Prediksi Lonjakan Penumpang Kereta Api Menggunakan Bayesian Extreme Value Theory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5002221079-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Peningkatan mobilitas masyarakat menyebabkan jumlah penumpang kereta api terus bertambah, terutama pada periode libur panjang, Lebaran, serta Natal dan Tahun Baru. Lonjakan penumpang yang bersifat ekstrem berpotensi menimbulkan risiko operasional dan sulit dianalisis menggunakan metode statistik yang berfokus pada nilai rata-rata. Oleh karena itu, diperlukan pendekatan yang mampu memodelkan kejadian ekstrem secara khusus. Tugas akhir ini bertujuan mengidentifikasi karakteristik tren data lonjakan penumpang ekstrem, mengestimasi parameter model Bayesian Extreme Value Theory (EVT), serta memperkirakan return level dan risiko lonjakan penumpang ekstrem pada stasiun-stasiun di Daerah Operasi (Daop) 8 Surabaya. Data yang digunakan adalah jumlah penumpang periode 2023–2025. Hasil eksplorasi data menunjukkan bahwa volume penumpang naik di Stasiun Gubeng kelas lokal eksekutif (A), Stasiun Babat kelas eksekutif (B), dan Stasiun Wlingi kelas eksekutif (C) memiliki karakteristik heavy tail yang mengindikasikan peluang terjadinya lonjakan penumpang ekstrem bernilai sangat besar. Nilai ekstrem ditentukan menggunakan pendekatan Block Maxima (BM) dengan distribusi Generalized Extreme Value (GEV) dan Peaks Over Threshold (POT) dengan distribusi Generalized Pareto Distribution (GPD), sedangkan estimasi parameter dilakukan secara Bayesian melalui simulasi Markov Chain Monte Carlo (MCMC). Hasil penelitian menunjukkan bahwa sebagian besar data ekstrem bersifat stasioner, kecuali data Stasiun C pada metode BM yang lebih sesuai dimodelkan menggunakan GEV nonstasioner. Estimasi parameter GPD pada seluruh stasiun mengikuti distribusi Pareto, sedangkan pada GEV, Stasiun A mengikuti distribusi Weibull serta Stasiun B dan Stasiun C mengikuti distribusi Fréchet. Estimasi return level menunjukkan peningkatan nilai ekstrem seiring bertambahnya periode ulang. Pada periode ulang 5 tahun, pendekatan BM menghasilkan return level sebesar 10.368, 2.770, dan 1.419 penumpang, sedangkan pendekatan POT menghasilkan 10.094, 2.261, dan 1.294 penumpang pada Stasiun A, B, dan C. Hasil tersebut menunjukkan potensi peningkatan risiko lonjakan penumpang ekstrem pada periode mendatang sehingga dapat menjadi dasar perencanaan kapasitas pelayanan dan mitigasi risiko operasional kereta api.
==================================================================================================================================
Increasing public mobility has resulted in continuous growth in railway passenger volumes, particularly during long holiday periods, Eid al-Fitr, Christmas, and the New Year. Extreme passenger surges may pose operational risks and are difficult to analyze using conventional statistical methods that focus on average behavior. Therefore, an approach specifically designed to model extreme events is required. This study aims to identify the trend characteristics of extreme passenger surge data, estimate the parameters of Bayesian Extreme Value Theory (EVT) models, and estimate the return levels and risks of extreme passenger surges at railway stations within Operational Area (Daop) 8 Surabaya. The dataset consists of monthly passenger volumes collected from 2023 to 2025. The results of the exploratory data analysis indicate that the upward passenger volumes at Gubeng Station in the local executive class (A), Babat Station in the executive class (B), and Wlingi Station in the executive class (C) exhibit heavy-tailed characteristics, indicating the potential for extremely large passenger surges. Extreme values were identified using the Block Maxima (BM) approach with the Generalized Extreme Value (GEV) distribution and the Peaks Over Threshold (POT) approach with the Generalized Pareto Distribution (GPD), while parameter estimation was performed within a Bayesian framework using Markov Chain Monte Carlo (MCMC) simulation. The results indicate that most extreme value series are stationary, except for Station C under the BM approach, which is better represented by a non-stationary GEV model. The GPD parameter estimates for all stations indicate Pareto-type behavior. In contrast, under the GEV model, Station A corresponds to the Weibull type, whereas Stations B and C correspond to the Fréchet type. The estimated return levels increase with the length of the return period. For a 5-year return period, the BM approach produced return levels of 10,368, 2,770, and 1,419 passengers, whereas the POT approach produced return levels of 10,094, 2,261, and 1,294 passengers for Stations A, B, and C, respectively. These findings suggest an increasing risk of extreme passenger surges in the future and provide valuable information for railway capacity planning and operational risk mitigation.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Bayesian Extreme Value Theory, Generalized Extreme Value, Generalized Pareto Distribution, Lonjakan Penumpang, Risiko Operasional. ============================================================ Bayesian Extreme Value Theory, Generalized Extreme Value, Generalized Pareto Distribution, Railway Passenger Surge, Operational Risk
Subjects:	Q Science > QA Mathematics > QA274.7 Markov processes--Mathematical models. Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA279.5 Bayesian statistical decision theory.
Divisions:	Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User:	Renata Najiah Efandy Putri
Date Deposited:	31 Jul 2026 10:12
Last Modified:	31 Jul 2026 10:12
URI:	http://repository.its.ac.id/id/eprint/140669

Actions (login required)

View Item