Please use this identifier to cite or link to this item:
Full metadata record
DC FieldValueLanguage
dc.contributor.authorNg, Michael Mun Hoeen_US
dc.identifier.citationNg, M. M. H. (2024). Parking availability prediction using regression methods and machine learning. Final Year Project (FYP), Nanyang Technological University, Singapore.
dc.description.abstractThis paper explores the application of regression methods and machine learning techniques for predicting parking availability, a crucial aspect of urban infrastructure management. The motivation behind this research lies in the increasingly pressing need to address urban congestion, a multifaceted issue with significant economic, environmental, and social implications. Urban areas around the world are experiencing unprecedented growth, leading to intensified strain on transportation systems, particularly concerning parking availability. Parking prediction is crucial due to its direct impact on urban transportation efficiency. Inadequate availability leads to wasted time, frustration, and increased congestion as drivers search for spots, worsening pollution and health issues. Additionally, accurate prediction aids urban planning by informing decisions on infrastructure, zoning, and policies, thus fostering sustainable, resilient cities and enhancing residents' quality of life. Existing approaches to parking prediction often rely on simplistic models or manual data collection methods, leading to limited accuracy and scalability. Traditional methods such as policy implementation and park choice modeling, while valuable, may not fully leverage the wealth of data available in the digital age. Furthermore, they may lack the sophistication required to capture the complex interplay of factors influencing parking availability, such as time of day, type of car park, location of car park, etc. Multiple regression models such as Linear Regression, Tweedie Regression, XGBoost Regression, and CatBoost Regression were employed and evaluated, with the Tweedie Regression model emerging as the most effective in terms of predictive accuracy, as indicated by favorable Mean Squared Error (MSE) and R-squared values. Neural network models such as the Random Forest Tree and Neural Network model were also trained to predict parking availability, with the Random Forest Tree outperforming the Neural Network model in accuracy, shown by the favorable MSE and R-squared values. However, the Tweedie Regression model still outperforms the best Machine Learning model, Random Forest Tree, and is used in the model selection and deployment. The selected model was implemented using Flask, a lightweight web application framework in Python, enabling real-time parking predictions. This allowed us to generate basic predictions on a webpage using HTTP requests. Additionally, a conceptual app design was proposed to visualize parking availability and aid users in planning their parking activities. The findings highlight the potential of machine learning in optimizing urban parking management systems, paving the way for more efficient and sustainable transportation solutions in urban environments.en_US
dc.publisherNanyang Technological Universityen_US
dc.titleParking availability prediction using regression methods and machine learningen_US
dc.typeFinal Year Project (FYP)en_US
dc.contributor.supervisorWang Zhiweien_US
dc.contributor.schoolSchool of Civil and Environmental Engineeringen_US
dc.description.degreeBachelor's degreeen_US
dc.subject.keywordsParking predictionen_US
dc.subject.keywordsRegression methodsen_US
dc.subject.keywordsMachine learningen_US
item.fulltextWith Fulltext-
Appears in Collections:CEE Student Reports (FYP/IA/PA/PI)
Files in This Item:
File Description SizeFormat 
  Restricted Access
1.3 MBAdobe PDFView/Open

Page view(s)

Updated on Jul 22, 2024


Updated on Jul 22, 2024

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.