Accurate Bus Arrival Time from Linear and Non-Linear Route Parameters Using Hybrid Predictors

Nowadays, public transportation plays a major role in various countries. Utilizing public transport is one of the important factors which indicates the quality of transportation department. The essential features of a quality public transport is providing precise information on bus departure and arrival schedules at bus stations. Accurate arrival time can be more helpful for the passengers to schedule their journey more comfortably. In this work, a new technique is presented to accurately predict bus arrival time at bus stops. Hybrid predictors are used in this work to perform the prediction process based on the route parameters. Route parameters are divided into linear and nonlinear to perform an independent prediction process. An extreme learning machine is used to perform prediction from nonlinear parameters. Also, a Support Vector Machine (SVM) predictor for processing linear route parameters. The proposed technique is tested on a local route and finds highly accurate result when compared to conventional techniques. This work achieved higher performance 0.05 of MAE, 5.5 of MAPE and 96.33 % of accuracy.