Resumen

Currently, intelligent systems are being implemented to meet the needs of different applications. In this case, in the agricultural field there are intelligent irrigation systems that, from agroclimatic stations, define actions for the irrigation of crops. The classic methods of irrigation are based on the switching on and off of the irrigation actuators, taking into account only the agroclimatic variables, leaving aside the farmer. In the present project, a decision support system for intelligent irrigation in an experimental garden was developed from the Agro-sensor device (project prior to this work) in order to generate a database of proposed irrigation rules. Subsequently, in the Python language through the Scikit-learn library, 4 supervised classifiers were implemented, among which are SGDClassifier, MLPClassifier, Adaboost and Gradientboosting, which were trained and validated with the data registered in the database, demonstrating among the results that Aboost was the classifier with the best performance, obtaining an F1-score score of 84.35%.