Traffic flow study is an essential application that helps city planners to develop and optimize a transportation network. These data can indicate congested areas of a transportation network and suggest plausible solutions. With the rise of intelligent transportation system (ITS) in modern traffic flow studies, traffic data are aggregated in real-time and it is more cost efficient than before. Due to the nature of the real-time intelligent system, the data are utilized in applications such as travel flow prediction to help facilitate the traffic. Other systems such as Google Maps, Bing Maps and Waze have utilized personal smartphones to collect rea-time data to attain even more accurate travel time estimates. However, it is very costly to maintain the infrastructure such as Google Maps and the collection of personal location data is a very sensitive topic in today’s climate where privacy is a major concern. In this project, I’m exploring the viability of using public resources, using the historical traffic data from the Washington State Department of Transportation and the National Weather Service to conduct a traffic flow case study using recurrent neural network to forecast traffic flow.
• Machine learning • Time series analysis • Recurrent Neural Network (RNN) • Long short-term memory (LSTM)
Intelligent transportation systems (ITS), Traffic flow, Travel time prediction, Washington State Department of Transportation (WSDOT), SR-405
Raw data isn't included in this repository. You can find the raw data from the WSDOT. You can find the results from the report.