A roadmap to solve two stage stochastic problems implementing scenario reduction for agricultural production planning under uncertainty

Leonardo Talero-Sarmiento, Juan David Márquez González, Henry Lamos-Díaz

Research output: EventsScientific eventspeer-review

Abstract

Decision-making in Agricultural Production Planning faces challenges due to the heterogeneous and uncertain sources, including but not limited to environmental conditions. (i) Temperature, rainfall, soil characteristics, and humidity. (ii) Non-constant factors in production such as hand-labor capacities, technology adoption barriers, finance availability, or low mechanization and automatization. (iii) Dynamic Sociopolitical context involving policies for or against rural development. Nevertheless, the farmer must decide over a defined decision horizon even if there is a lack of information. Consequently, this work's main objective is to provide a well-structured and detailed description of methods for solving optimization models under uncertainty scenarios; promptly, the Stochastic Optimization model and its two-stage strategy implementing forecasting scenarios. This work addresses a three-stage approach, starting with a section focusing on a (1) brief analysis of modeling trends for optimal decision-making during agricultural production planning under uncertainty, highlighting the (*) sources of uncertainty, and (*) comparing modeling strategies. The following section presents a detailed and illustrative (2) roadmap to develop a two-stage model from scratch, covering concepts of (*) large-scale optimization techniques such as decomposition and (*) processes to model the uncertainty of the parameters covering (*) scenario generation, including its probability estimation. (3) The final section shows an illustrative case to apply the Roadmap addressing a Two-stage stochastic problem. The case study uses the ARIMA modeling and backward reduction through the Kantorovich distance cluster strategy to represent the parameter trajectories, solving the deterministic equivalent formulation stochastic problem using the Benders decomposition technique. This work's critical contribution is that the Roadmap provides a valuable guide for practitioners, students, and researchers who need to model agricultural production planning under uncertain conditions providing a well-structured case study based on data reported (SP).
Original languageSpanish (Colombia)
Pages131
Number of pages142
StatePublished - 5 Aug 2022
EventCongreso Internacional Industria y Organizaciones - Universidad Nacional de Colombia, Bogotá, Colombia
Duration: 4 Aug 20225 Aug 2022
Conference number: XI

Conference

ConferenceCongreso Internacional Industria y Organizaciones
Abbreviated titleCIIO 2022
Country/TerritoryColombia
CityBogotá
Period4/08/225/08/22

Research Areas UNAB

  • Modelamiento matemático y estadística aplicada

Cite this