Rekomendasi Lokasi Halte Bus Dalam Konsep 15-Minute City Di Surabaya Berbasis Fitur Graf

Amanda, Adhira Riyanti (2025) Rekomendasi Lokasi Halte Bus Dalam Konsep 15-Minute City Di Surabaya Berbasis Fitur Graf. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025211102_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB)

Abstract

Kota Surabaya sebagai pusat perekonomian dan bagian dari kawasan metropolitan di Indonesia mengalami lonjakan urbanisasi yang signifikan. Hal tersebut berdampak pada meningkatnya kebutuhan transportasi publik. Meskipun telah tersedia moda transportasi publik seperti Suroboyo Bus, Trans Semanggi Suroboyo, dan Wira Wiri Suroboyo, cakupan aksesibilitas terhadap Tempat Perhentian Kendaraan Penumpang Umum (TPKPU) masih belum merata. Penelitian ini bertujuan untuk membangun sistem prediksi lokasi TPKPU yang lebih inklusif dan berkelanjutan dengan mengadopsi pendekatan 15-Minute City, yakni konsep perencanaan kota dimana kebutuhan dasar masyarakat dapat diakses dalam waktu 15 menit dengan berjalan kaki atau bersepeda. Metode yang digunakan mencakup rekayasa fitur graf berbasis lima ukuran centrality (degree, betweenness, closeness, eigenvector, dan pagerank) dari jaringan jalan di Surabaya, serta penerapan pseudo-labelling dengan algoritma LightGBM untuk mengatasi keterbatasan data berlabel. Selanjutnya, hasil klasifikasi dikombinasikan menggunakan metode Soft Voting dari beberapa model machine learning seperti XGBoost, LightGBM, AdaBoost, Random Forest, dan Decision Tree yang sudah dioptimalkan terlebih dahulu melalui proses hyperparameter tuning. Evaluasi menunjukkan bahwa model kombinasi Decision Tree dan LightGBM menjadi yang terbaik untuk moda Suroboyo Bus-Trans Semanggi Suroboyo, sementara kombinasi LightGBM dan Random Forest unggul untuk moda Wira-Wiri Suroboyo. Kinerja model-model diuji pada empat studi kasus (area sekolah, rumah sakit, masjid, dan universitas). Pada studi kasus Pelajar Sekolah, model Suroboyo Bus mencatat F1-Score sebesar 90,00%, sedangkan model Wira-Wiri mencapai 76,06%. Pada studi kasus Pasien Rumah Sakit, model Suroboyo Bus mencapai 94,12%, dan model Wira-Wiri memperoleh 92,68%. Pada studi kasus Masjid Ampel, model Suroboyo Bus memperoleh F1-Score sebesar 100,00%, dan model Wira-Wiri 92,68%. Sementara itu, pada studi kasus Universitas, model Suroboyo Bus mencatat F1-Score sebesar 70,00%, dan model Wira-Wiri sebesar 94,44%. Hasil prediksi dari studi kasus ini selanjutnya dilakukan post-processing menggunakan sistem inferensi fuzzy untuk menghasilkan persentase kelayakan setiap titik sebagai lokasi TPKPU potensial.
==================================================================================================================================
Surabaya, as an economic center and part of a metropolitan area in Indonesia, is experiencing a significant surge in urbanization. This has led to an increased demand for public transportation. Although public transportation modes such as Suroboyo Bus, Trans Semanggi Suroboyo, and Wira Wiri Suroboyo are already available, accessibility to Bus Stops (TPKPU) remains uneven. This study aims to develop a predictive system for TPKPU locations that is more inclusive and sustainable by adopting the 15-Minute City approach—a concept of urban planning where basic needs can be accessed within 15 minutes by walking or cycling. The method involves graph-based feature engineering using five centrality measures (Degree, betweenness, closeness, eigenvector, and pagerank) from Surabaya's road network, as well as the application of pseudo-labelling with the LightGBM algorithm to address the limitation of labeled data. The classification results are then combined using the Soft Voting method across several machine learning models, including XGBoost, LightGBM, AdaBoost, Random Forest, and Decision Tree, all of which were optimized through hyperparameter tuning. Evaluation results show that the combination of Decision Tree and LightGBM performs best for the Suroboyo Bus–Trans Semanggi Suroboyo mode, while the combination of LightGBM and Random Forest performs best for the Wira Wiri mode. Model performance was tested across four case studies: school, hospital, mosque, and university areas. In the school case, the Suroboyo Bus model achieved an F1-Score of 90,00%, and the Wira Wiri model reached 76.06%. In the hospital case, Suroboyo Bus scored 94.12%, and Wira Wiri scored 92.68%. In the mosque case, Suroboyo Bus achieved a perfect score of 100,00%, while Wira Wiri reached 92.68%. In the university case, the Suroboyo Bus model recorded 70,00%, and the Wira Wiri model achieved 94.44%. e model predictions are then post-processed using a fuzzy inference system to generate the probability of each location being recommended as a potential TPKPU site.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	15-Minute City, Graph Centrality, Fuzzy Inference System, Machine Learning, Pseudo-labelling, Soft Voting, Tempat Perhentian Bus.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Adhira Riyanti Amanda
Date Deposited:	30 Jul 2025 09:55
Last Modified:	30 Jul 2025 09:55
URI:	http://repository.its.ac.id/id/eprint/123847

Actions (login required)

View Item