Predicting Solar Water Pump Adoption Among Smallholder Farmers in Ethiopia: A Socioeconomic and Policy Modeling Approach

Abstract

This research assessed and predicted small‐scale farming and irrigation practices in Ethiopia’s Amhara region for the deployment of solar water‐lifting systems. A structured survey was conducted with 200 households, 100 from Dangisheta (irrigated) and 100 from Zegie Ura (rain‐fed). Data were analyzed using statistical and machine learning models, including logistic regression, random forest, and support vector classification, to predict adoption probability and identify key determinants. The random forest model achieved 76.7% accuracy, outperforming logistic regression and SVM (both at 73.3%). Results indicate that 48% of nonadopters cited high cost as the primary barrier; simulations showed that removing this barrier could double adoption to 95.5% (from a 47% baseline). Demographic factors such as secondary education increased adoption, while each additional year of age decreased by 3.2%. The study contributes to integrating survey‐based socioeconomic data with predictive modeling and policy simulations, offering a scalable framework to inform solar pump adoption strategies in underserved rural areas. Solar pumps were found to be the most viable and cost‐effective solution for sustainable irrigation in Ethiopia’s diverse agro‐climatic conditions.