Agriculture is one of the main contributors to carbon emissions. Understanding different processes involved in farming and estimating the carbon emissions in each step can help in accurately calculating the carbon factor and support in optimizing and reducing the carbon emissions. Potato is a popular food product cultivated across the world. Potato farming involves several processes such as preparing the land, using fertilizers and manures, irrigation, and plowing, and all these steps have been contemplated to generate carbon emissions. This article investigates the steps involved in potato cultivation as a case study and generates standardized features related to carbon emissions in each step. Different machine learning and deep learning algorithms are used to model these standard features. This research predicts the carbon emission using different regression models such as random Forest, multiple linear regression, lasso regression, K-Nearest Neighbour, and neural network regression and finally compares them based on the metrics of root-mean-square error (RMSE) and R2. The results show that all the models have comparable performance with a R2 score very close to 1 and very low RMSE. The novelty of the work is in introducing standard features for modeling carbon emissions in agriculture which help to streamline different farming datasets even across different crops. |
*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.