Skripsi

ANALISIS PERBANDINGAN ALGORITMA MACHINE LEARNING UNTUK PREDIKSI KETERLAMBATAN KEBERANGKATAN LIGHT RAIL TRANSIT

---

This research was conducted because departure delays on the Light Rail Transit can reduce passenger comfort and satisfaction. Therefore, a predictive model that can estimate delays accurately is needed. The purpose of this study is to implement, compare, and determine the best machine learning algorithm for predicting LRT departure delays in Canberra. The dataset used consists of static and real time GTFS data for the period 2020–2022, which went through several stages including data cleaning, feature extraction, categorical encoding, outlier handling, normalization, and chronological split consisting of 80% training data and 20% testing data. Three algorithms were applied, namely Random Forest Regressor, Extreme Gradient Boosting, and Support Vector Regressor, and the training process was carried out using Bayesian Optimization. The evaluation results show that XGB achieves the best performance, with an R² value of 0.97, MAE of 2.08, and RMSE of 4.30 on the test data, as well as an R² score of 1.00, MAE of 0.97, and RMSE of 1.57 on the training data. Feature importance analysis indicates that arrivalDelay, actual_departure_minute, and stop_headsign are the most influential features. These results demonstrate that the machine learning approach is effective in accurately predicting LRT departure delays. Keywords : GTFS, departure delay, Light Rail Transit, machine learning, Random Forest Regressor, Support Vector Regressor, XGBoost

---

Availability

Detail Information

Series Title

-

Call Number

T1931772026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xiv, 81hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

006.320 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Komputer
Perbandingan Algoritma

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment