Analysis and forecast of emission values in Turkish transportation systems

Today, more than 95% of the energy consumed for transportation worldwide consists of vehicles using fossil fuels. Emissions from energy sources used for the transport sector account for approximately 20% of worldwide emissions. This situation becomes a major threat in terms of environmental pollution. A large part of environmental pollution consists of air pollution. Harmful emissions such as carbon dioxide (CO2), hydrocarbons, nitrogen oxides (NOx) and sulphurous components resulting from the work done with energy sources lead to depletion of the ozone layer and bad consequences for our country and the world, such as global warming. In this study, the analysis and estimation of the emission values originating from the transportation systems in Turkey are made. Artificial neural networks (ANN), multivariate linear regression (MLR), fuzzy logic and logarithmic regression methods are used for these analyzes. Linear, Interaction and Purequadratic methods of MLR method are applied. The analyzes are evaluated on the basis of the correlation coefficient (R), mean of squares of errors (MSE) and mean percent error (MPE) criteria. As a result, the R value obtained from ANN was 0.9809, the MSE value was 8050363681418.75 and the MPE value was 4.1555%. The R value of the linear method of the MLR model was 0. 9855, the MSE value was 4793845004079.31, and the MPE value was 3.7989%. The R value of the interaction method was 1.0000, the MSE value was 0.000000004839, and the MPE value was %0.0000000001672. The R value of the Purequadratic method was 0.9900, the MSE value was 3314943364316.79 and the MPE value was 2.5781%. The R value of the logarithmic regression method was 0.9872, the MSE value was 4236601138306.92, and the MPE value was 3.4057%. The R value of the fuzzy logic method was obtained as 0.9855, the MSE value was 4895753910316,73 and the MPE value was 4.9089%. As a result of the estimations obtained, the method that gives the best result is observed as the interaction method of the multivariate linear regression method.