Colloquium on Transportation Systems Engineering and Management CTR, CED, NIT Calicut, India, May 12-13, 2014. Paper ID: 173 PATTERN BASED SPATIAL FORMULATION FOR BUS TRAVEL TIME PREDICTION UNDER MIXED TRAFFIC CONDITIONS B. Anil Kumar 1 ., Lelitha Vanajakshi 2 and Shankar C. Subramanian 3 1 Graduate Student, Dept. of Civil Engineering, IIT Madras, raghava547@gmail.com. 2 Associate Professor, Dept. of Civil Engineering, IIT Madras, lelitha@iitm.ac.in. (Corresponding Author) 3 Associate Professor, Dept. of Engineering Design, IIT Madras, shankarram@iitm.ac.in ABSTRACT In recent times, congestion has become a serious problem in Indian cities due to rapid changes in urbanization. There is a need to explore better traffic operation and management systems to overcome congestion related problems such as delays and pollution. In this regard, attracting more people towards public transport is one of the option to reduce the congestion levels. To attract more people, the public transit should provide quality services to passengers. This can be achieved in one way by providing real-time information to the passengers about bus arrival details at bus stops. The effectiveness of such an information provided to passengers highly depends on the prediction method used, which in turns depends on the input data used in the prediction method. Thus, identifying correct input and incorporating them in the prediction model is important. Using the identified significant patterns in the data, a model based algorithm was developed to predict next bus travel time. The model is tested for a selected MTC bus route, 5C, which connects Taramani and Parry’s corner in Chennai city, India. The performance the proposed algorithm showed a clear improvement in prediction accuracy when compared with a prediction method that uses only previous two trips as input to predict next bus travel time. Keywords: bus travel time; significant input; Kalman Filtering Technique; statistical test. 1. INTRODUCTION The prediction of bus travel time and providing such information about bus arrival time to passengers waiting at bus stops accurately is a very important aspect in Advanced Public Transportation Systems (APTS). It will be useful to passengers for reducing their anxieties and waiting time at bus stops, or to make reasonable travel arrangement before making a trip. However, the information provided to passengers should be reliable; otherwise customers may reject the system due to lack of reliability (Schweiger, 2003). The effectiveness of such information provided to passengers greatly depends on the prediction technique used, which in turns depends on the input data used in the prediction method. Thus, identifying most significant input data and incorporating them into in prediction methods will hopefully improve the accuracy of the performance. In the present study, bus travel time prediction using a model based approach was attempted. The most significant data to be used in the prediction method was identified by using statistical analysis results obtained from Kumar et al. (2013)