A Machine Learning approach for Short-Term Traffic Flow Prediction - MSc Research Project Data Analytics

Sep 17, 2026

Uploaded by Noah

Traffic congestion creates complex economic, environmental, and health impacts, creating a need for accurate short-term forecasting to support traffic management. This MSc research predicts short-term traffic flow using statistical, machine learning, and deep learning models on the PEMS-08 dataset (San Bernardino), covering July–August 2016 at five-minute intervals. Three case studies compare LSTM, CNN, RNN, KNN, Random Forest, Gradient Boosting, and Decision Tree, showing Random Forest, Decision Tree, and KNN as top performers (R2 0.99–0.97) with very low RMSE and MAE. K-fold validation supports reliability and robustness. Results inform policymakers and enable proactive strategies toward sustainable urban transport by addressing dynamic traffic patterns.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages26

Reading Time1 h 6 min

AccessFree to download

Abstract

1 Introduction

A Machine Learning approach for Short-Term Traffic Flow Prediction MSc Research Project Data Analytics Aaditya Ravindra Gajendragadkar Student ID: 22158758 School of Computing National College of Ireland Supervisor: Furqan Rustam National College of Ireland Project Submission Sheet School of Computing | Student Name: | Aaditya Ravindra Gajendragadkar | | --- | --- | | Student ID: | 22158758 | | Programme: | Data Analytics | | Year: | 2023 | | Module: | MSc Research Project | | Supervisor: | Furqan Rustam | | Submission Due Date: | 14/12/2023 | | Project Title: | A Machine Learning approach for Short-Term Traffic Flow Prediction | | Word Count: | 9000 | | Page Count: | 24 | I hereby certify that the information contained in this (my submission) is information pertaining to research I conducted for this project. All information other than my own contribution will be fully referenced and listed in the relevant bibliography section at the rear of the project. ALL internet material must be referenced in the bibliography section. Students are required to use the Referencing Standard specified in the report template. To use other author’s written or electronic work is illegal (plagiarism) and may result in disciplinary action. | Signature: | Aaditya Ravindra Gajendragadkar | | --- | --- | | Date: | 30th January 2024 | PLEASE READ THE FOLLOWING INSTRUCTIONS AND CHECKLIST: | Attach a completed copy of this sheet to each project (including multiple copies) . | □ | | --- | --- | | Attach a Moodle submission receipt of the online project submission, to each project (including multiple copies) . | □ | | You must ensure that you retain a HARD COPY of the project, both foryour own reference and in case a project is lost or mislaid. It is not sufficient to keep a copy on computer. | □ | Assignments that are submitted to the Programme Coordinator office must be placed into the assignment box located outside the office. | Office Use Only | | | --- | --- | | Signature: | | | Date: | | | Penalty Applied (if applicable): | | A Machine Learning approach for Short-Term Traffic Flow Prediction Aaditya Ravindra Gajendragadkar 22158758 Abstract Traffic congestion presents a complex problem, impacting not just the convenience of commuters, but also imposing significant economic and environmental consequences. While various models have been employed in traffic prediction, there remains a need for accurate, robust, and practical solutions to aid traffic managersin understanding patterns, reducing congestion, and optimizing traffic management strategies. This research focuses on predicting short-term traffic flow using statistical, machine learning, and deep learning models on the PEMS-08 Dataset from San Bernardino, covering July to August 2016 in five-minute intervals. In three case studies, deep learning models of LSTM, CNN, RNN, and machine learning models like KNN, Random Forest, Gradient Boosting, and Decision Tree demonstrate commendable performance, especially Random Forest, Decision Tree, and KNN, outshining others and making them first choice with R2 values of 0.99,0.98 and 0.97 respectively with extremely low RMSE and MAE values. Deep learning models LSTM, CNN, and RNN follow closely with R2 values of 0.958,0.957, and 0.91 respectively but with slightly higher RMSE and MAE, and statistical models of SARIMA and ARIMA Performed well with R2 of 0.95 and 0.90 but with an extremely high RMSE and MAE values. K cross-validation is performed on each machine learning and deep learning model that confirms the model’s performance, robustness, and reliability. This research offers valuable insights to policymakers, presenting optimal models for developing proactive strategies. The findings contribute to fostering sustainable and efficient urban transportation systems by addressing dynamic traffic patterns. 1 Introduction Today in the world of rapid urbanization every smart city faces a massive challenge that br

FAQs

What problem does this research address?

It addresses traffic congestion and the need for accurate, practical short-term traffic flow prediction to support traffic management and reduce congestion effects.

Which dataset and time scope are used for model training and evaluation?

The models use the PEMS-08 dataset from San Bernardino, covering July to August 2016 in five-minute intervals.

Which models achieved the best predictive performance?

Random Forest, Decision Tree, and KNN performed best, with R2 values around 0.99, 0.98, and 0.97 and very low RMSE and MAE.

Sep 17, 2026

Uploaded by Noah