Paper Title
Gap Acceptance Analysis at an Unsignalised Intersection using Machine Learning Techniques

Abstract
As the vehicular population of Hyderabad is increasing rapidly with compound annual growth rate (CAGR) of 13.84%. Deficiency in infrastructure to accommodate the present traffic is causing an increase in travel time. This is exaggerated by much delays especially at Unsignalised intersections. In this study, analysis of gap acceptance at an unsignalised intersection is evaluated and parameters like Approach speed, type of vehicle, Waiting time, Number of vehicles objecting subject vehicle were considered. To analyze the gap acceptance at an unsignalised intersection, Machine Learning Techniques like Artificial neural network (ANN) and Support vector regression (SVR) models were developed. ANN model is found to be the best fit model than SVR. Critical gap during peak hour using Harder’s model and Raff’s model were also calculated. To evaluate the performance of the calibrated models, measures like Mean Absolute Deviation (MAD), Root Mean Square Error (RMSE), Mean Squared Prediction Error (MSPE), Mean Absolute Percentage Error (MAPE), Schawarz’s Bayesian Criterior (SBC) were used. Keywords - Gap appectance, Unsignalised intersection, ANN, SVR, Critical gap