Development of a System for Analyzing and Predicting Agronomic Conditions to Optimize Banana Harvest Yield at the "La Fortuna" Banana Plantation

Abstract

Banana cultivation is one of the main pillars of the Ecuadorian economy; however, in many farms productivity is still limited by traditional agronomic monitoring practices, fragmented records, and inefficient administrative processes. In this context, the objective of this research was to implement a comprehensive web-based system for the analysis and prediction of agronomic conditions in order to optimize crop yield at the "La Fortuna" farm, located in Valencia canton. The study followed a mixed, applied, and descriptive approach and was supported by the Scrum framework to organize the technological development process. During the diagnostic stage, 15 strategic stakeholders from the farm participated, enabling the identification of operational needs, the standardization of 18 critical variables climatic, edaphic, nutritional, phytosanitary, productive, and management and the integration of information into a single platform. A Ridge Regression model was implemented for yield prediction, reaching a coefficient of determination of R² = 0.70 and a Mean Absolute Percentage Error of 17.9%, values that demonstrate adequate predictive capacity to support operational planning. The platform reduced by 93% the administrative time required to generate reports measured through real-time chronometry before and after implementation and achieved a user acceptance rate of 96.7%, assessed through a five-point Likert scale. The integration of data analytics, digital traceability, and automated prediction strengthens decision-making and promotes more precise, timely, and sustainable agronomic management.