APPLICATION OF REGRESSION EQUATIONS IN LIMNOLOGICAL RESEARCHES: ADVANTAGES OF USING ARTIFICIAL NEURAL NETWORKS

Abstract

In the present paper the accuracy of regression models prediction of some important parameters of lake ecosystems (primary production, chlorophyll a concentration, zooplankton and zoobenthos biomass) is analyzed on the basis of literature data. It was shown that the prediction accuracy, measured as the mean absolute percentage error (MAPE), in almost all cases reaches 60-100%, what does not allow these models to be used for expert assessments of the ecosystem parameters of lakes. Using the literary data, multiple regression models were generated on the base of artificial neural network technology. Verification of the accuracy of these models was performed on independent data that were not used to build this model. Neural network regression models turned out to be more accurate – their mean absolute percentage error did not exceed 25%. Thus, in our opinion, the advantage of using regression neural network models in limnological studies is very perspective.