Prediction of Water Quality Parameters of River Periyar Using Regression Models

The recent trends in urbanization and industrialization have resulted in the deterioration of water quality in both developed and developing countries. India has recently become a water-stressed country, with a rise in demand for freshwater and rising levels of pollution. Parameters like pH, dissolved oxygen (DO), and biochemical oxygen demand (BOD) can be effectively used to determine the water quality of various water bodies. The water quality parameters for seven years of pH, BOD, and DO, from the various sample stations of river Periyar, were used to predict the expected values of these parameters, using various regression models like Linear, Least Absolute Shrinkage and Selection Operator (LASSO), Bayesian, and Stochastic Gradient Descent (SGD). A comparison, based on mean square error, of the predictions using the various models, has been presented. The linear regression model provided the best estimation for the pH, while the SGD model gave the best estimate for the BOD and DO. An Internet of Things (IoT) system is also proposed which can accrue real-time daily data to help improve the prediction accuracy of the models.