Comparative Effectiveness of Some Approaches to Extracting Most Informative Factors Influencing Algae Bioproductivity

Water eutrophication is a worldwide environmental problem. Understanding the mechanisms of the process help to prevent and remediate such eutrophication. The paper presents researches of the dependence of chlorophyll a concentration in phytoplankton on a number of different physicochemical factors on the basis of long-term observation data in the Kakhovka reservoir of Dnipro River. Four modeling methods were applied in this work: linear regression, two sorting-out GMDH algorithms (complete search combinatorial algorithm COMBI and directional search correlation-based sorting algorithm with regressors rating analysis CAR) and the LASSO. Main stress is laid on the possibility of automatic (more objective) discovering the most informative factors influencing algae bioproductivity.