Predicting total alkalinity in saline water using machine learning: A case study with RapidMiner

Abstract

This study investigates the use of machine learning models to predict total alkalinity (TA) based on chloride concentration (Cl^-), pH and temperature. Utilizing RapidMiner's Auto Mode, six machine learning models were applied to a dataset of 111 water samples from the Nhà Bè River. The models' performances were evaluated using Root Mean Square Error (RMSE) and R² metrics, with the Generalized Linear Model (GLM), Support Vector Machine (SVM) and Deep Learning models identified as the top performers. Correlation and coefficient analyses revealed that Cl^- had the most significant impact on TA prediction, followed by temperature and pH. These findings underscore the effectiveness of machine learning in water quality monitoring, presenting a cost-effective alternative to traditional chemical analysis methods.