A predictive water quality model for the estimation of water pollution :

Abstract

Water is an essential element to both human life and the entire ecosystem .This water pollution caused by different activities such as agricultural activities ,indiscriminate waste disposal among others .River Malaba located in Eastern Uganda which is transboundary water body shared by Uganda and Kenya and whose water flows to lake Kyoga is at risk of pollution as a result of the human activities taking place along the river such agricultural practices, sand mining , industrial effluent and urban waste from Malaba town among others this therefore puts the beneficiary population at risk of water pollution adverse effects like waterborne diseases such as diarrhea ,typhoid among others. This study developed a machine learning predictive water quality model basing on Random Forest classifier algorithm having produced an accuracy of 87.6% on the following physio-chemical parameters PH, Dissolved Oxygen, Nitrates, Phosphates, Color, Turbidity and total coliform after model testing .This model is therefore developed to assist in real-time control of future water quality changes thus simplifying judgement of the degree of water pollution of water pollution hence improving management level of river Malaba and also providing data to policy makers and environment management teams around the river acting as basis for early warning.