Modelling of the trip generation with artificial neural network

There are multiple variables that affect trip generation. As well as the effects of these variables may be individually, they may be effective all together. The aim of this study is to modelling the trip generation with Artificial Neural Networks (ANN) by using the numbers of population, gross domestic product and registered cars from these variables influencing the trip generation and to compare with the regression results.Due to the fact that there are many uncertainties affecting trip generation, works have been carried out with the method of ANN and benefitted from MATLAB programme.The research has been carried out to modelling the trip generations to occur on state highways in 81 provinces for the year of 2009 by using 2007-2008 data, in 8 chosen provinces for the year of 2009 by using 2007-2008 data and in 8 provinces by using 2009 data.As a result of these three applications, the estimations by ANN have given better results comparing to classic regression method. In the first application R-square distribution of 0.914 for ANN LOG, R-square distribution of 0.928 for ANN TAN and R-square distribution of 0.560 for MRA have been obtained. In the second application R-square distribution of 0.991 for ANN LOG, R-square distribution of 0.970 for ANN TAN and R-square distribution of 0.946 for MRA have been obtained. In the third application application R-square distribution of 0.942 for ANN LOG, R-square distribution of 0.953 for ANN TAN and R-square distribution of 0.922 for MRA have been obtained.Three variables influencing the trip generation have been evaluated together in the research carried out. It is possible to develop the research by increasing the number of variables.