1. Introduction
Regression analyses are commonly carried out in the Environmental research in order to establish relationships among variables and predict the response or determine the cause of certain continuous variable of interest. There isa wide variety of methods intended for resolving regression tasks, including linear regression (LR), multilayer perceptron (MLP) and model trees (M5P). Bayesian networks (BNs) belong to the so-called probabilistic graphical models and their application has exponentially increased in environmental sciences, although scarcely to solve regression tasks, in the recent years [1]. In this study we compare the aforementioned 3 techniques with BNs in terms of their accuracy, aiming at showing that BNs are able to deal with regression tasks and perform proper