Application of Artificial Intelligence to Predict CO2 Emissions: Critical Step towards Sustainable Environment

Prediction of carbon dioxide (CO2) emissions is a critical step towards sustainable environment. In any country, increasing the amount CO2 an indicator increase in environmental pollution. this regard, current study applied three powerful and effective artificial intelligence tools, namely, feed-forward neural network (FFNN), adaptive network-based fuzzy inference system (ANFIS) long short-term memory (LSTM), to forecast yearly Saudi Arabia up year 2030. The data were collected from “Our World Data” website, which offers measurements years 1936 2020 for every country on globe. However, only concerned with related Arabia. Due some missing data, considered 1954 2020. 67 samples divided into 2 subsets training testing optimal ratio 70:30, respectively. effect different input combinations prediction accuracy was also studied. inputs combined form six groups predict next value past values. group that contained addition as temporal index found be best one. For all models, performance accuracies assessed using root mean squared errors (RMSEs) coefficient determination (R2). Every model trained until smallest RMSE reached throughout entire run. FFNN, ANFIS LSTM, averages RMSEs 19.78, 20.89505 15.42295, respectively, while R2 0.990985, 0.98875 0.9945, individually emission. To benefit powers (AI) final forecasted average (ensemble) models’ outputs. assess forecasting accuracy, ensemble validated new measurement 2021, calculated percentage error 6.8675% 93.1325%, implies highly accurate. Moreover, resulting curve ensembled models showed rate expected decrease 9.4976 million tonnes per based period 1954–2020 6.1707 2020–2030. Therefore, finding work could possibly help policymakers take correct wise decisions regarding issue not near future but far future.