Feature Extraction from Trajectories for Transportation Mode Detection in Smart City using Machine Learning

Nowadays, with the advancement of technology and the importance of intelligent transportation in the smart city, predicting and identifying the use of transportation modes is the primary and essential part of intelligent transportation such as traffic control, travel demand analysis, and transportation planning. With the pervasiveness of positioning devices as well as the pervasive use of smartphones, recorded location routes are an economical and rapid approach to identifying modes of transport. In this study, using situational data recorded by the Microsoft Research Asia (Geolife) and extracting dynamic characteristics, walking, biking, use of bus, driving, and use of train are predicted. The proposed approach to using machine learning classifications includes support vector machine classification, random forest classification, gradient boosting classification, and extreme gradient boosting classification. Among the models, the extreme gradient boosting model was able to predict transport modes with higher accuracy by obtaining 95.18% accuracy and less time complexity.