FORECASTING GREENHOUSE GAS EMISSIONS IN THE INDUSTRIAL PRODUCTION OF THE REPUBLIC OF KAZAKHSTAN

Abstract

Excessive greenhouse gas (GHG) emissions are an environmental problem. Studies to determine cost-effective ways to reduce GHG emissions have revealed the need to model the dynamics of emissions of carbon dioxide, nitrous oxide, methane, and other gases. In this study, the calculation of CO2 equivalent emissions from industrial processes and production in the territory of the Republic of Kazakhstan was carried out. When forecasting, the data provided by the UN Framework Convention on Climate Change were used. To predict CO2 emissions from industrial production, tools for analysis and forecasting of time series were used: Prophet method, Cluster analysis of k-means time series, modern versions of ARIMA algorithms, exponential smoothing methods, and linear regression. This study presents comparative simulation results based on a baseline scenario with no action until 2045.This study compares four models to suggest an effective one for future CO2 emission forecasting. The accuracy comparison is conducted using various error measures, with the mean absolute percentage error (MAPE) chosen as the metric for comparison.