Modeling dissolved oxygen concentration using machine learning techniques with dimensionality reduction approach.

Area of Science:

• Environmental Science
• Water Quality Monitoring
• Predictive Modeling

Background:

• Dissolved oxygen (DO) is critical for aquatic life and ecosystem health.
• Accurate prediction of DO levels is essential for effective water resource management.
• Existing monitoring methods require enhancement with predictive capabilities.

Purpose of the Study:

• To develop an accurate prediction model for dissolved oxygen (DO) concentrations in aquatic systems.
• To evaluate the performance of Random Forest (RF) and multilayer perceptron (MLP) algorithms for DO prediction.
• To assess the impact of dimensionality reduction on model performance.

Main Methods:

• Utilized Random Forest (RF) and multilayer perceptron (MLP) algorithms for regression modeling.
• Applied wrapper feature selection for dimensionality reduction of the dataset.
• Evaluated model performance using Pearson correlation coefficient, Mean Absolute Error (MAE), and Root Mean Square Error (RMSE).

Main Results:

• The RF regressor outperformed the MLP algorithm in predicting DO concentrations.
• Dimensionality reduction significantly reduced estimation error deviation for both models.
• The RF regressor achieved the best performance with a reduced variable dataset, showing a Pearson correlation coefficient of 0.8052.

Conclusions:

• Random Forest (RF) is a robust regressor for predicting DO concentrations.
• Dimensionality reduction positively impacts the accuracy of DO prediction models.
• Feature selection enhances the predictive power of machine learning models for water quality parameters.