Water Quality Index Forecasting Using Machine Learning

Abstract

The forecasting of water quality in the Tumkur district, Karnataka state, India, was conducted using several machine learning methods, such as support vector machines, regression tree, linear regression, and neural network. The Water Quality Index (WQI) was calculated using factors like total hardness, pH, alkalinity, turbidity, chloride, dissolved solids, and conductivity. A ratio of 80:20 was used to divide the dataset into two groups for the purposes of validating and training the models. Over the past few years, water quality has been negatively impacted by multiple pollutants, making it crucial to predict and model water quality for effective pollution reduction. Advanced machine learning methods were devised for this studyto forecast the WQI. The models' effectiveness was assessed using a variety of statistical and visual assessment techniques. Among the models used, Support Vector Machine and Linear Regression exhibited superior performance, with R2 value 0.96 and 0.99 respectively for the train and test data sets. The implementation of AdaBoost, an ensemble model, for forecasting WQI also yielded excellent results, achieving R2 values of 1 and 0.91 for the training and testing data, respectively