Answers are grounded in this paper and cite their sources.
Smart agricultural water-management technologies can improve irrigation efficiency, but implementation depends partly on farmers’ willingness to use them. This study evaluated an explainable artificial intelligence framework for predicting stated adoption intention, defined as continued conventional water supply (0) versus intended smart-system use (1). The analysis included 468 observations from 92 governance-program participants in a single paddy-dominated reservoir irrigation command area. Logistic regression, elastic-net logistic regression, random forest, radial-basis-function support vector machine, and XGBoost were compared across five cumulative feature blocks, from farmer and farm characteristics to perceived compatibility, using repeated nested respondent-grouped cross-validation. Adding irrigation-management perceptions increased mean AUROC by 0.208 across algorithms. Among the cumulative specifications, the complete feature set achieved the highest mean AUROC for each algorithm; XGBoost achieved an AUROC of 0.897 and a Brier score of 0.098. However, compatibility used directly as an ordinal score achieved a comparable AUROC of 0.898. SHAP identified compatibility as the dominant contributor to fitted-model predictions, followed by technology awareness and irrigation-schedule evaluation. Shared measurement and conceptual proximity limit interpretation of compatibility’s predictive contribution. The findings may inform farmer consultation, while external validation and prospective evaluation are needed to establish operational usefulness and the effectiveness of any subsequent support activities.
No takes yet. Share an insight, caveat, or question.
Da Ye Kim (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: