Constructing regression models by identifying the relative influence of biotic and abiotic factors on plankton biomass in the USA

Plankton are fundamentally important to aquatic ecosystems. They are responsible for constituting the base of the food chain as well as driving important aquatic cycles. They are also notable during plankton blooms, which can be dangerous to subsurface marine life. Thus, regression models able to capture the patterns of plankton biomass would be very profound. This study correlates 44 different water-quality and water-chemistry variables with plankton biomass to identify statistically significance. 5 variables were identified at the 90\% confidence level. 7 regression models are then trained and evaluated on both interpolation and extrapolation each using all variables and only significant variables. All models performed better using only significant variables in both problems. Decision trees performed best for interpolation, and ridge regression performed best for extrapolation.