EDGE COMPUTING AND ADVANCED DATA ANALYTICS IN MONITORING CHEMICAL POLLUTION EFFECTS ON MARINE LIFE

Abstract

This paper aims to understand the feasibility of implementing edge computing and analytical tools to detect chemical contamination and its impact on organisms. Through K-Means clustering, Random forest, Support Vector Machine (SVM) and Long Short-Term Memory (LSTM) network we made a study on the collected pollution data from marine sensors. K-Means Clustering identified three distinct pollution clusters: Cluster 1 can be characterized as low, while in Cluster 2 the level of pollution is significantly higher and belongs to moderate level, Cluster 3 can be characterized with high level of pollution. On the basis of results, Random Forest has accuracy of 85%, precision 82%, recall 87% and F1- score 84%. With SVM, accuracy was 88%, precison was 85%, recall was 90% and F1-Score was 87 %. It was revealed that LSTM networks were useful in making good prediction of future pollution levels with an MSE of 0. 05 and root mean squared error (RMSE) of 0. 22. The incorporation of edge computing helped to make real-time data analytics and analysis, thus increasing the effectiveness of pollutants’ identification and control. This research focuses on the exploration of how advanced analytics and edge computing can support the especially the external environment monitoring and decision making.