A machine learning algorithm to explore the impact of agricultural land-use types on CO2 emissions in Vietnam in the period 1990-2019.

The paper deals with an application of machine learning algorithms to examine the impact of agricultural land-use types on CO₂ emissions in Vietnam during the period 1990-2019. A four-layer Artificial Neural Network (ANN) model was employed to analyze the relationship between 21 agricultural land-use types (LUTs) and CO₂ emissions. The selected LUTs cover major agricultural products, including crops, meat, and vegetables. The results indicate that most agricultural LUTs are positively associated with CO₂ emissions, including bananas, dry beans, cabbages, cashew nuts in shell, fresh cassava, raw or retted jute, cauliflowers and broccoli, dry chilies and peppers, raw cinnamon and cinnamon-tree flowers, coconuts in shell, green coffee, groundnuts excluding shelled, fresh hen eggs in shell, watermelons, tea leaves, and sweet potatoes. By contrast, fresh or chilled horse meat, unmanufactured tobacco, soya beans, sesame seed, and rice show negative associations with CO₂ emissions. These findings provide empirical evidence to support policymakers in designing targeted strategies for reducing greenhouse gas (GHG) emissions from agricultural production in Vietnam.