Forecasting Carbon Dioxide Emissions Of Bahrain Using Singular Spectrum Analysis And ARIMA Hybrid Model

Abstract

Predicting the economic implications of gas emissions and their repercussions is criti-cal to policymakers, especially given the current increasing trend in volume. Therefore, study on gas emission prediction is required. A hybrid model is proposed for forecasting CO2 emissions of Bahrain (BH) in this study. Singular Spectrum Analysis (SSA) and Auto Regressive Integrated Moving Average are the two techniques that make up the hybrid model (ARIMA). In this model, the time series of CO2 emissions are first divided into a number of sub-series corresponding to some tendentious and oscillation (periodic or quasi-periodic) components and noise by using SSA. Each sub-series is then predicted individually through an appropriate ARIMA model. Finally, a correction procedure is carried out for the sum of the prediction results to ensure that the superposed residual is a pure random series. The result of forecasting obtained by using hybrid SSA-ARIMA was compared with ARIMA to assess its superiority. While, the application of SSA was to improve the accuracy of forecasting of CO2 emissions in BH. The data used in this study are the annual time series data in three different periods from 1990-2018, 2000-2018 and 2003-2018 for CO2 emissions in BH. The Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE) criteria are used to assess each forecasting methods level of forecasting accuracy. The results indicated that the SSA-ARIMA method is superior to the ARIMA model as the best forecasting method for CO2 emis-sions.