Performance Assessment of Urban Streets Addressing Improvement Issues for Automobile Mode of Transport

This paper focusses on modeling automobile drivers’ response patterns to assess urban-street service quality in developing countries. Several quality-of-service attributes affecting drivers’ riding quality were investigated, from 102 urban street segments under widely varying geometric and traffic conditions. Traffic volume, effective road width, travel speed, pavement condition, on-street parking turnover, land use, hindrance due to public transits, non-motorized vehicles, and encounters are found to significantly affect drivers’ comfort levels. Two novel artificial intelligence techniques, that is, artificial neural network (ANN) and functional linked artificial neural network (FLANN) were applied to predict automobile drivers’ level of satisfaction scores ( ALOS_score). The prediction performance of developed models is assessed in terms of various statistical parameters of a modified rank index. Bayesian regularization neural network has given the best fitted model in both training and testing data sets among the ANN models. However, application of the FLANN model shows better prediction performance in the present context, as no hidden layer exists. All input layer neurons are directly linked with output layer neurons with a lesser number of connections, which is advantageous over ANN in reducing accumulated error. The result shows 73% of studied segments are offering service category “C” or below. Sensitivity analyses reported that pavement condition is the most important variable, with relative importance of 26.78%, to influence drivers’ riding quality. Similarly, other parameters were ranked in decreasing order of their relative importance, which will help highway authorities to prioritize budgets of future investments for improving service quality.