This paper analyzes the interactive effects of freight and passenger transport on Turkey’s energy consumption(TEC)in the trans-portation sector to guide policymakers on improving energy efficiency.From 1975 to 2019,T...This paper analyzes the interactive effects of freight and passenger transport on Turkey’s energy consumption(TEC)in the trans-portation sector to guide policymakers on improving energy efficiency.From 1975 to 2019,TEC,freight transport(FT)and passenger transit(PT)demand rose significantly-by factors of 5.02,4.82 and 4.97,respectively.Accurate forecasting of future TEC,FT and PT is crucial for informed infrastructure decisions.The study employs six machine learning algorithms,including two time series,two ensemble and two neural network models,to predict 342 models related to TEC,FT and PT.Performance metrics such as R^(2),MAPE,RMSE,nRMSE and RPE showed that the controlled ARIMA(CARIMA)and swapped ARIMA(SARIMA)models were most effective.A hy-brid CARIMA-SARIMA-LGM model outperformed single-parameter models.For forecasting until 2050,26 ARIMA and four exponential smoothing models were developed,with ARIMA outperforming the latter.Predictions indicated TEC,FT and PT would increase by factors of 2.02,1.73 and 1.81,respectively.The linear regression model found that FT contributed 22.08%and PT 31.54%to TEC be-tween 2020 and 2030.By 2050,contributions were 20.32%for FT and 28.41%for PT,with residual factors accounting for the remainder.Key influences on energy demand growth include infrastructure development,technology,policies,economic factors,fuel prices and consumer behaviour.展开更多
基金supported via funding from Prince Sattam Bin Abdulaziz University with the project number(Grant No.PSAU/2025/R/1446).
文摘This paper analyzes the interactive effects of freight and passenger transport on Turkey’s energy consumption(TEC)in the trans-portation sector to guide policymakers on improving energy efficiency.From 1975 to 2019,TEC,freight transport(FT)and passenger transit(PT)demand rose significantly-by factors of 5.02,4.82 and 4.97,respectively.Accurate forecasting of future TEC,FT and PT is crucial for informed infrastructure decisions.The study employs six machine learning algorithms,including two time series,two ensemble and two neural network models,to predict 342 models related to TEC,FT and PT.Performance metrics such as R^(2),MAPE,RMSE,nRMSE and RPE showed that the controlled ARIMA(CARIMA)and swapped ARIMA(SARIMA)models were most effective.A hy-brid CARIMA-SARIMA-LGM model outperformed single-parameter models.For forecasting until 2050,26 ARIMA and four exponential smoothing models were developed,with ARIMA outperforming the latter.Predictions indicated TEC,FT and PT would increase by factors of 2.02,1.73 and 1.81,respectively.The linear regression model found that FT contributed 22.08%and PT 31.54%to TEC be-tween 2020 and 2030.By 2050,contributions were 20.32%for FT and 28.41%for PT,with residual factors accounting for the remainder.Key influences on energy demand growth include infrastructure development,technology,policies,economic factors,fuel prices and consumer behaviour.
