The amount of greenhouse gases in the atmosphere is increasing day by day. This increase is caused primarily by global warming, resulting in numerous negative effects. Predicting future greenhouse gas emissions can be encouraging, especially in terms of decision makers and sectors with a share in CO2 emissions, to reduce this emission and to seek alternative sources. Time series is the name in the literature of the data obtained regularly at regular intervals on the time plane, and the processes that examine how these series are analyzed are called time series analysis. The study used a data set containing 55 years of data from the World Bank database containing the greenhouse gas emissions (CO2 equivalent) values of Turkey. It is aimed to obtain useful patterns from this data set with artificial neural networks and exponential smoothing methods. The data set, which was converted to time series format for analysis, was then divided into two parts as training and test data. The training data in the time series type was analyzed using Holt's linear trend model, which is based on the exponential smoothing method, and artificial neural networks (NSA), which is one of the sub-branches of artificial intelligence. As a result of these analyses, prediction models were obtained based on training and test data. Assessment metrics such as RMSE, MAPE were obtained to evaluate the models with the predictions of ANN and Holt's linear trend method. According to these values, two models were compared and it was determined that the model with the least error rate was ANN. According to the findings obtained in the study, YSA has RMSE value of 0.1607 and it has a much lower error rate compared to Holt's linear trend method. After finding that the YSA would make more accurate predictions, estimates were obtained by 2021 using the model proposed by this method. The Model estimated Turkey's greenhouse gas equivalent to CO2 emissions in 2021 at 366,3972 million tons. Another result seen in the research is that CO2 emissions follow a fluctuating course but tend to increase in general.
Publication Date : August 31, 2020
|APA||Özhan, E . (2020). Yapay Sinir Ağları ve Üstel Düzleştirme Yöntemi ile Türkiye’deki CO2 Emisyonunun Zaman Serisi ile Tahmini . Avrupa Bilim ve Teknoloji Dergisi , (19) , 282-289 . DOI: 10.31590/ejosat.705666|