Dynamic Selection of Machine Learning Models For The Ambiental Analysis of Agroecosystems

Abstract

Artificial Intelligence (AI), and in particular Machine Learning (ML), provides powerful tools for extracting knowledge from complex data and supporting well-founded, evidence-based decision-making. In this context, while Decision Support Systems (DSS) are essential for analyzing large volumes of data, a persistent challenge lies in making them sufficiently flexible and adaptable to a wide range of problems. This work presents the design and implementation of a web-based DSS, focused on biological datasets, which leverages dynamic model selection in ML. The system automatically adapts to different datasets by selecting the most appropriate model based on the structure and quality of the available data. The methodology follows a modular approach, comprising several stages—from database upload and target variable selection to the prediction of new entries for decision support. The work concludes with a case study using a dataset involving chemical and biological aspects related to pesticide residue levels in honeybee hives and their environmental impact.