Since several decades, monitoring and control of water quality continue to attract a great deal of interest and, indeed, there have been a number of developments in this subject. However, the application of machine learning algorithms for water quality has significantly increased which have explored in number in the early of last years. Hereafter, we explore the capability of the various machine learning models, i.e., adaptive boosting (AdaBoost), Categorical Boosting (CatBoost), Extreme Gradient Boosting (XGBoost), and Light Gradient-Boosting Machine (LightGBM), for predicting critical water quality variables namely, the Phycocyanin fluorescence water in situ (PC). The PC variable was modelled using various water quality variables, i.e., water pH, water specific conductance (SC), water temperature (Tw), dissolved oxygen concentration (DO), and discharge (Q). The performances of the four machine learning models were compared based on numerical performances, i.e., the root mean squared error (RMSE), the mean absolute error (MAE), the coefficient of correlation (R), and the Nash-Sutcliffe efficiency (NSE), and graphical visualization. In depth, analysis and comparison between the models were done by highlighting some important conclusions. Results obtained revealed that, PC can be predicted with high precision and accuracies using all models and the XGBoost exhibited the high performances with R ≈ 0.922, NSE ≈ 0.849, RMSE ≈ 0.271 mg/L, and MAE ≈ 0.191 mg/L.