Dissolved oxygen is one of the parameters of river water quality for which the accurate estimation is critical and of significant importance for many environmental applications. The continuous monitoring at high frequency (1 hour) of the water quality of the Sebou at the Mechra Bel Ksiri gauging station began in June 2020 using a multiparameter probe YSI (EXO2) and sensors. Real-time water quality data were collected in the Sebou River for Dissolved Oxygen, pH, temperature, electrical conductivity, turbidity, chlorides, and nitrates. Water level measurements were recorded at the same frequency by the radar. In this study, the high-frequency measurement highlights nycthemeral cycles at the day scale. A sine function model and three machine learning techniques: XGB Regressor, Random Forest Regressor, and Linear Regression, were used to develop data-driven models using temperature, nitrate levels, pH, and electrical conductivity to reproduce the daily fluctuations of dissolved oxygen. Each machine learning model was calibrated for each input scenario using 80% of the data obtained between June and November 2020 and validated using the remaining 20%. The evaluation simulations quality included the calculation of the error indices between simulations and observations (MAE and RMSE) and coefficient of determination (R²). The prediction based on the three machine learning techniques gave good results for XGB Regressor, Linear Regression, and Random Forest Regressor with R 2 values of 0.93, 0.89, and 0.88, respectively, while the sine function was slightly over-predicting the daily cycle of dissolved oxygen (R² =0.80) and indicated a relatively high saturation rate. The error indices were very low, with MAE and RMSE ranging in 0.05-0.06 and 0.06-0.08 respectively, which suggests satisfactory models performance.