The combination of light, temperature, conductivity, oxygen, nitrates and metals form a group of important stress environmental factors that influence on biodiversity. The last ones, the metals, several of these parameters are associ- ated with agricultural and farming activities. Therefore, increased agricultural ac- tivity could lead to disruption in biodiversity equilibrium, especially on ecosys- tem like Lake Prespa. Discovering the right influencing factor on diatom biodi- versity is the task that this paper aims to shade a light on. We plan to achieve this by using state of the art methods for machine learning. Since several metal pa- rameters are influencing the diatom biodiversity, multi-target regression tree method is used. We investigate different strategies and we pick the best model(s) based on the experimental evaluation. The obtained models reveal that Na and Mg are the most influencing factors on the diatom biodiversity. Based on these results, further research based on this method for other abiotic stress factors could be made.
Multi-Target Regression Models, Over-fitting, Ensembles, Metal Parameters, Biodiversity