Estimating the Travel Demand for Road Traffic using Regression and Neural Network Methods (Case study: Qom-Tehran Freeway)

Travel is one of the most important aspects of human life and is one of the most important periodic (or non-periodic) activities. For this reason, over the years, various ways have been devised to meet this need of communities. Similar many industrial systems, the transportation system are affected by supply and demand relations, and any action in this area should be planned according to existing relationships and future trends of supply and demand. In this research, using 3 Models: simple linear regression, multivariate regression and multi-layered perceptron neural network models for forecasting traffic demand of Qom-Tehran (freeway) axis was studied. The data used in this research include the Iranian Statistical Center, statistical manuals and Qom traffic information. Independent variable in multivariate regression and neural network including population, working population, income level and simple linear regression model of population. The results of this study show that Pearson correlation between variables considered in neural network methods, multivariate regression, linear regression was 0.995, 0. 933, and 0,723 respectively, and the success rate of each of these models in estimated dependent variable (travel demand) were 0.99, 0.885, and 0.541, respectively. Comparison of methods has shown that the neural network method has the highest correlation and accuracy and simple linear regression method has the least correlation and precision in demand estimation.