OSE-WQI: An Optimized Stacked Ensemble Classifier to aid Water Quality Assessment

Water is an essential resource for human life. Safe and pure water is an important component of the ecosystem. Freshwater covers about 2.5% of the earth’s surface, and only 1% of it is usable. River water has a significant proportion of freshwater which is used for various purposes. However, excessive exploitation and inappropriate use of water resources have led to water pollution. The degraded water quality can cause transmission of diseases and it cannot be used for drinking, agricultural and industrial use. Analyzing the water quality has become one of the prime aspects of water management and monitoring. In this work, machine learning techniques are adopted to automate the process of water quality assessment. The complete process is divided into two stages. In the first stage correlation among water parameters is identified and water quality factors are forecasted. During this process, the regression method is applied to forecast the missing water quality parameters. These forecasted parameters along with the original parameters are then used to formulate a Water Quality Index (WQI) which is further used to categorize the water quality using a stacked ensemble classification approach. The proposed approach is implemented on Yamuna River’s data collected from various sampling locations In the Delhi region. The experimental analysis shows that the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) for predicting Biological Oxygen Demand (BOD) and Chemical Oxygen Demand (COD) are 0.4132, 0.1707, and 0.5134, 6.0588 respectively For the next stage classification scenario, a comparative analysis shows that the proposed approach achieves an overall performance as 95.833, 91.66, 92.31, and 92.05% in terms of accuracy, precision, recall and F-score, respectively.