Estimation of water quality index in Zohreh River using principal component analysis and artificial intelligence models

This research explored the root causes of hidden pollution and key factors affecting spatial changes, as well as identifying the best inputs for water quality modeling. The study used principal component analysis (PCA), artificial neural network models (MLP), gene expression programming (GEP), and support vector machine (SVM) to achieve its objectives. The dataset included 11 different parameters collected monthly over 10 water years (2012-2021) from the Zohreh River, Iran. Initially, PCA was applied to reduce parameters and calculate the Water Quality Index (WQI). Two input models (parameters before and after PCA) were then created using artificial intelligence to determine the most accurate model for predicting the WQI. The Kaiser-Meyer-Olkin measure (KMO) was 0.6524, indicating the dataset was suitable for factor analysis. Bartlett's sphericity test was also significant at the 0.05 alpha level. PCA identified five significant principal components, explaining 70.66% of the total variance. The combined SVM and PCA model showed the best prediction ability, with an R² of 0.889, RMSE of 0.052, and MAE of 0.038.