Advantages of Machine Learning in Bus Transport Analysis - Bus On-Time Performance Study

Aug 3, 2026

Supervised machine learning provides a data-driven way to emulate learning from past experience and translate it into predictive decision-making. This study applies supervised models to Tehran BRT bus punctuality, using publicly available datasets from 2020 to 2022 supplied by the Municipality of Tehran for training and testing. Models built with Python SciKit-Learn and Statsmodels achieve accurate on-time performance predictions and enable interpretation of influential factors, supporting actionable insights for improving route effectiveness and schedule adherence.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Abstract

Introduction

Machine Learning in Transportation

Problem Explanation

Dataset

Advantages of Machine Learning in Bus Transport Analysis Amirsadegh Roshanzamir M.Sc Graduated, Department of Electrical Engineering, Sharif University of Technology [roshanzamir_as@alum.shaif.edu](roshanzamir_as@alum.shaif.edu), ABSTRACT Supervised Machine Learning is an innovative method that aims to mimic human learning by using past experiences. In this study, we utilize supervised machine learning algorithms to analyze the factors that contribute to the punctuality of Tehran BRT bus system. We gather publicly available datasets of 2020 to 2022 from Municipality of Tehran to train and test our models. By employing various algorithms and leveraging Python's Sci Kit Learn and Stats Models libraries, we construct accurate models capable of predicting whether a bus route will meet the prescribed standards for on-time performance on any given day. Furthermore, we delve deeper into the decision-making process of each algorithm to determine the most influential factor it considers. This investigation allows us to uncover the key feature that significantly impacts the effectiveness of bus routes, providing valuable insights for improving their performance. KEYWORDS: Machine Learning, Decision Tree, Supervised Algorithm, Classifier Algorithm, Bus Transportation. I. INTRODUCTION The rise of the internet and technology has revolutionized the way we record and share information. We now have an overwhelming amount of data at our fingertips, making it nearly impossible for humans to extract meaningful insights efficiently. However, machines have emerged as valuable allies in this data-driven era, helping us make sense of the vast information available. While humans can often make accurate judgments, quantifying the decision-making process can be challenging. This is where supervised machine learning steps in. Supervised machine learning algorithms aim to provide a logical explanation for why something is the way it is. This approach closely resembles how humans learn. For example, when a child is shown pictures of kittens and told by their parent that those images represent kittens, the child gradually learns to correctly identify a kitten when they see one. Humans can classify new instances based on past experiences. Supervised machine learning seeks to replicate this behavior by manually labeling a subset of a dataset and using computers to recreate this experience-based learning process. Our goal is to apply established supervised machine learning techniques to solve a unique problem. II. MACHINE LEARNING IN TRANSPORTATION Machine learning algorithms have been widely used in the field of public transportation to address issues such as bus punctuality. These algorithms have been employed to predict variables like bus arrival times, travel durations, and dwell times. With advancements in data collection and storage technologies, machine learning has become an essential tool in analyzing the performance of public transit systems. For example, in Seoul, South Korea, a non-parametric regression method called nearest neighbor has been used to analyze bus travel time [2] . This algorithm considers historical and current path travel times of neighboring paths to predict bus travel time. Compared to the traditional plain historical average technique, the nearest neighbor algorithm has shown superior accuracy, especially during daytime when travel time fluctuations are more pronounced. Another aspect studied in public transportation is bus bunching, where buses tend to bunch together instead of maintaining a regular spacing. Factors such as the day of the week, bus dwell time, intersection delay, schedule deviation, bus spacing, and proximity to bus stops are used to model this problem [1]. Machine learning algorithms like Gene Expression Programming and Decision Tree have been employed to identify the most influential factors in bus bunching. The Decision Tree algorithm has emerged as the most accurate, with th

FAQs

How does this study use supervised machine learning for Tehran BRT punctuality?

The study trains supervised machine learning models on Tehran BRT data from 2020 to 2022 to predict whether bus routes meet prescribed on-time performance standards on a given day.

What dataset is used to train and test the models?

It uses publicly available datasets provided by the Municipality of Tehran, containing bus-recorded on-time performance data across routes, with modifications to incorporate location information.

Which tools and algorithms are used in the analysis?

The work employs Python libraries such as SciKit-Learn and Statsmodels, and it uses supervised classifiers including approaches like Decision Tree to identify influential factors for punctuality.

Aug 3, 2026

Upload by Olivia Brown