A Multi-Criteria Predictive Framework for Sustainable Tomato Irrigation

Abstract

This chapter presents a multi-criteria predictive framework designed to support sustainable tomato irrigation under increasing climate constraints. The work introduces baseline modeling that uses Random Forest, XGBoost, and MLP Regressor algorithms applied to three clean-irrigation datasets covering environmental, agro-physiological, and field conditions. Model performance is evaluated through R2, MAE, and RMSE, while the TOPSIS method ranks model–dataset combinations across multiple criteria. Although the datasets do not yet include treated wastewater variables, the resulting reference models establish a controlled benchmark for future integration of chemical and biological contaminant data. This chapter therefore provides the foundational architecture needed to extend predictive irrigation systems toward safe and sustainable treated wastewater reuse.