Machine Learning techniques for Hydroponic Cultures

Abstract: Hydroponics is an innovative agricultural technique that enables the cultivation of plants without the use of soil, providing controlled conditions for optimal plant growth. In recent years, machine leaming (ML) techniques have gained prominence in various domains, including agriculture, due to their ability to analyze large datasets and derive valuable insights: the combination of ML and the opportunity to control all the inputs in an hydroponic cultivation represents an invaluable chance to reduce resource requirements and increase the production in line with the constraints imposed by climate change. In this work, we tested four different machine learning models, namely, random forest (RF), support vector machine (SVM), extreme gradient boosting (XGB) and a neural network. These models were tested in two different scenarios considering two sets of variables. The first scenario is done considering all the features of the dataset while the second scenario is characterized only by the features that can be measured during the cultivation. The best result is obtained in the second scenario with extreme gradient boosting (XGB) that achieved a value of 8.37 for mean absolute error (MAE), 8.20 for mean bias error (MBE) and 13.16 for root mean square error (RMSE).