Özet:

Greenhouse gas emissions prevent our world's self-renewal capacity and cause ozone depletion, global warming and reduced food resources. In addition, greenhouse gases are the biggest factor that creates the ecological footprint. To make the world more livable and self-sufficient, the biocapacity fields and the ecological footprint must be in balance. In order to achieve this balance, the situation for the future of greenhouse gas emissions should be determined. In this study, the forecasting of greenhouse gas emissions for Turkey is carried out using machine learning algorithms, and the data set denominated greenhouse gas emissions of Turkey between the years 1967-2017. In order to test the success of the methods, the data set is first handled as a time series and then 10-fold cross-validation is applied to evaluate the results statistically. Long Short-Term Memory is determined as the best algorithm and in the test set evaluated as time series, root mean square error, mean absolute percentage error and the coefficient of determination of this algorithm are found as 0.25, 1.11, and 1.0 respectively. The model created with these successful results is used to estimate greenhouse gas emissions between 2018 and 2031. Forecasted emission values are at a high level compared to today, and necessary measures and activities to increase biomass should be carried out considering these values.

Anahtar Kelimeler: