Prediction of 5‐day biochemical oxygen demand in the Buriganga River of Bangladesh using novel hybrid machine learning algorithms

Abstract

Biochemical oxygen demand (BOD) is one of the most important variables indicating stream pollution with a severe condition of organic loading and maintaining aquatic life in ecosystems. Advanced monitoring techniques such as machine learning (ML) methods have been developed for an accurate, reliable, and cost-effective prediction of BOD. This study investigated the effectiveness of four stand-alone ML algorithms, namely, artificial neural network (ANN), support vector machine (SVM), random forest (RF), and gradient boosting machine (GBM), and six novel hybrid algorithms, namely, RF-SVM, ANN-SVM, GBM-SVM, RF-ANN, GBM-ANN, and RF-GBM, in predicting BOD of the Buriganga river system of Bangladesh. The feature importance analysis of RF algorithm indicated that chemical oxygen demand (COD), total dissolved solids (TDS), conductivity, total solids (TS), suspended solids (SS), and turbidity are the most influential parameters for predicting BOD 5 . The significance of this study is the application of the novel hybrid models that resulted in higher prediction success; RF-SVM with the highest R 2 value (0.908). The employed novel hybrid ML models can be particularly useful for efficient and systematic data management, water pollution control, and prevention in developing countries such as Bangladesh. Practitioner Points• Investigated the efficiency of four stand-alone and six novel hybrid ML models for predicting BOD in a river of Bangladesh.• The significance of this study is the application of the six novel hybrid models that resulted in higher prediction success.• The best three prediction models were RF-SVM, ANN-SVM, and GBM-SVM with a prediction success of 91%, 89.6%, and 88.8% respectively.• ML models indicated COD, conductivity, TDS, TS, SS, and turbidity as the most influential variables for predicting BOD.• The novel hybrid models can be useful for developing countries for efficient systematic data management, pollution control, and prevention strategies.Nabila Nafsin is a member of the Water Environment Federation (WEF).

Prediction of total organic carbon and E. coli in rivers within the Milwaukee River basin using machine learning methods

“…Several studies [7][8][9][10] developed ML models for event detection in water distribution systems. Other studies [11][12][13][14][15][16] investigated the performances of different ML models for predicting water quality parameters of natural source water. ML models have also been developed to predict the water quality index and water quality class.…”

Estimation and mapping of water quality parameters using satellite images: a case study of Two Rivers Dam, Kenya

“…However, there was a slight variation between Landsat-predicted and in situ TSS since Landsat estimated the TSS values with a coefficient of variation of less than 10%. The model results can be compared with the results from the study by Nafsin & Li (2022) which investigated the effectiveness of four stand-alone machine learning (ML) algorithms and six novel hybrid algorithms in predicting the 5-day BOD of Buriganga River, Bangladesh. The Random Forest-Support Vector Machine (RF-SVM), Artificial Neural Network-Support Vector Machine (ANN-SVM), and Gradient Boosting Machine-Support Vector Machine (GBM-SVM) achieved high prediction accuracies of 91, 89.6, and 88.8%, respectively.…”