Predictive Modelling for Healthy Marine Life and Ecosystem using Ensemble Techniques

Abstract

Marine life analysis is the study of physical conditions of ocean body such as temperature, pressure, alkalinity, dissolved oxygen and carbon content by carefully analyzing the optimal measurements of these parameters for marine living organisms to thrive safely underwater. This analysis gives a clear idea about how marine living conditions change with respect to the change in temperature, pressure and other similar parameters. Such analysis can be used to find the optimal parameters for a given water body by deploying machine learning algorithms to predict these range of specifications. In this paper, a comparative study on machine learning algorithms is done to predict the ocean parameters from the dataset provided by NOAA National Centers for Environmental Information. This dataset consists of data on pressure, dissolved inorganic carbon, total alkalinity, water temperature, salinity and dissolved oxygen, with respect to depth, obtained from the Canadian Beaufort Sea. This data is used efficiently in predicting all other parameters, given one among them. This methodology is also effectual, since knowing one of the parameters using any sensor can easily predict the optimal conditions for the other parameters, hence turning out to be more efficacious as a system.