Online Traffic Speed Forecasting Based on Multi-Periodicity Gaussian Process Models

• Main Authors: Alexander, 陳智文
• Other Authors: Hsing-Kuo Pao
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/76944453632717842478

碩士 === 國立臺灣科技大學 === 資訊工程系 === 104 === Intelligent Transportation Systems (ITS) has been developed to aid drivers and other road-users to make a better travel decision. In recent years, many researches have been conducted in this field. Being one kind of time-series data, traffic data also follows the general aspects of time-series, which are periodicity and trend. This research highlights the periodicity aspects while also considers more specific aspects such as feature correlations and unexpected patterns. In fact, thanks to the periodicity of the traffic data, most drivers can tell how the traffic state will be on the road they are familiar with. However, this is not the case for drivers who are not familiar with the road. Here we aim to provide an approach which is able to consider both periodicity and unexpected patterns of the traffic data. We choose Gaussian Process Regression as our main model since it has the ability to explore implicit relationships between multiple variables in the traffic data.