Feature selection method for traffic prediction

• Main Author: Zhang, Guokuan
• Other Authors: Justin Dauwels
• Format: Theses and Dissertations
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/68665
• Institution: Nanyang Technological University

Short-term traffic prediction has been one of the most essential parts of most Intelligent Transportation Systems (ITS). As a main target in traffic prediction, improving the prediction accuracy has always been the research focus. Feature selection provides an optimal pre-processing step in traffic prediction. In this dissertation report, several pre-processing steps were investigated and applied, such as finding spatial features, extracting links data, using Local Fisher Discriminant Analysis (LFDA), etc. After the feature selection, Support Vector Machine (SVM) was introduced on the new data space for traffic forecasting. Some experiments were conducted on the dataset of Singapore transportation. And the experimental results have shown the effectiveness and efficiency of the combined prediction strategy.