Authors: Komathy Karuppanan
Addresses: Funded Research Laboratory, Department of Computer Science and Engineering, Academy of Maritime Education and Training, Chennai, 600097, India
Abstract: Quality of marine water has a direct impact on the evolution of the ocean ecosystem, which aggravates the survival of marine organisms and animals. Additionally, the ballast water, carried in ships drawn from the open sea/ocean, may lead to bio-invasion if the quality of the water is unchecked and released into another sea/ocean. The ballast water treatment methods may sometime alter the physicochemical properties, which might be harmful to receiving water region. The objective of this paper is, therefore, to propose a new method for the estimation of marine water quality from the physiochemical and biological properties of the marine water and then a suitable computational architecture supporting machine learning techniques to assist in classifying and predicting the marine water quality. The proposed framework has evaluated various classification models to select the best-fit algorithm for this application through model training and optimising. The finalised model called stacking classifierwas then recommended for ballast water quality prediction with 100% accuracy, which could be deployed prior to the ballast water exchange.
Keywords: data modelling; machine learning; water quality index; water quality; model optimising; model training; prediction model; marine water quality classification; prediction accuracy; correlation; ensemble learning.
International Journal of Water, 2022 Vol.15 No.1, pp.21 - 38
Received: 21 Jun 2022
Received in revised form: 14 Nov 2022
Accepted: 15 Nov 2022
Published online: 11 May 2023 *