Travel time model using gps data and machine learning for bus information systems

Abstract

Public Transportation modes are prevalent and extensively utilised means of transportation for commuters. Road congestion, bus crew issues, malfunctions, and miscellaneous factors impede buses from adhering to schedules. As a result, it is becoming problematic for commuters to arrange their travel plans confidently. Intelligent transportation systems use Global Positioning System (GPS) technology and data analytics to accurately predict real-time travel information and improve traveller and operator experience. The research gap is the unavailability of standardised techniques for mass travel time predictions using standardised analytical methods. However, recent research has focused on developing accurate travel-time models employing machine learning algorithms. Predictive models rely on past data gathered through GPS systems. The study uses the GPS data of public buses in Central Province, Sri Lanka, one thousand buses have been fitted with GPS units since 2019 (7 million of data). Realtime and historical data that were gathered through GPS units can be used to develop machine learning-based models to predict bus or passenger transport information accurately. The study analysed available data using Microsoft Azure, Statistical, Time Series and Machine algorithms for performance accuracy with lower error rates on predictions used for comparison purposes.