Estimating Travel Speed of Yangon Road Network Using GPS Data and Machine Learning Techniques

Travel speed estimation at the city level is the essential problem advantageous to many forms of urban transport applications. This paper proposes a travel speed estimation of road network from the GPS data of public transportation system by analyzing trajectory data using machine learning techniques and simple linear and non-linear equations. Map matching algorithms and KNN model are used to project trajectory on to the map and to determine the start and end points of road segments, respectively. The average travel speed of every road segments along the bus route are estimated by using historical data and live GPS data of running buses. Features related to traffic delay are extracted from the Place of Interest (POI) near the road segment. The speed factor matrix is generated based on the output of Travel Speed Estimation Model and traffic delay related features.