Evaluation of the impact of tariff price on load distribution and network traffic by data mining method

The transportation tariff is always one of the serious challenges that the stakeholders of the transportation market are dealing with, which always has a direct effect on traffic and capacity. In this study, after performing various corrections on the data and categorizing and removing outliers, the most suitable model was built and implemented in the experiment design software as a white box, which is accurate and easy to use. In addition to normal pricing, the proposed method can be used as an auxiliary method to improve the pricing process. Now, if the effects of variables such as inflation rate, load distribution, etc. follow a stable trend, the above method can be directly used in pricing. Railways should determine the pricing strategy that only has a service perspective or that has a revenue perspective as well. The results showed that the neural network model has high accuracy, but it takes about two hours to use this method every time. Deep learning system was also implemented and reported higher accuracy. Goodness of fit was reported as 0.9925, but running time was slightly longer. The next method that was implemented was various regression models, the goodness of fit of which reached 0.427 in the best option due to the inappropriateness of the data, which is a very low value and is not acceptable. Therefore, the use of these models was not approved.