Modeling and production forecast for transitory crop planning

We address the tactical planning of transient crops, with a focus on potato, quinoa, sweet potato, and cassava. These crops exhibit short, finite life cycles, requiring accurate alignment between planting and harvesting windows. We propose a bi-objective mixed-integer linear programming model in which the primary production parameter is determined through a data-driven, learning-based approach. The result is an integrated decision-support framework that combines supervised machine learning with mathematical optimization for agricultural planning. The scope includes scheduling planting and harvesting operations and resource allocation, emphasizing the temporal and spatial coordination of decisions across multiple regions under conflicting objectives: maximizing total production while minimizing logistical complexity. The mathematical formulation incorporates neighborhood structures and logical constraints to enforce the temporal continuity of operations. The proposed methodology is validated through a case study involving the Peruvian agroecological regions of Arequipa, Cusco, and Puno. Computational experiments are conducted on instances ranging from small-scale scenarios with 90 plots to large-scale scenarios with up to 3000 plots. The results indicate that the proposed approach effectively identifies non-dominated solutions, balancing production maximization with integrated planning for planting, harvesting, and machinery allocation. The proposed framework has the potential to serve as a decision-support tool for efficient farm management while handling high-dimensional binary decision spaces.