Probabilistic Forecasting of Carbon Emissions in Thailand using Bayesian Deep Neural Networks

The global concern regarding climate change has centered around reducing CO2 emissions. Precise carbon emission forecasts are vital, as they allow policymakers, businesses, and individuals to make informed decisions about reducing emissions and transitioning to a low-carbon economy. This paper introduces a long-term probabilistic forecast of Thailand's carbon emissions using Bayesian deep neural networks. The model uses Gross Domestic Product, population, energy consumption, oil consumption, electricity generation and renewable energy generation as input parameters. The model is evaluated using the historical dataset from the Energy Policy and Planning Office, Ministry of Energy, Thailand, from 1990 to 2023. The impacts of increasing renewable energy generation, energy consumption, and oil consumption on carbon emissions are also evaluated.