Machine learning algorithms to predict the catalytic reduction performance of eco-toxic nitrophenols and azo dyes contaminants (Invited Article)

Removing hazardous substances like azo dyes and nitrophenols from drinking water is essential for maintaining human health since these substances occur naturally in the environment. This research study used machine learning techniques to estimate the catalytic reduction performance of environmentally hazardous nitrophenols and azodyes pollutants. The catalyst PdO-NiO was used to eliminate contaminants in the water, including 4-nitrophenol (4-NP), 2,4-dinitrophenol (DNP), 2,4,6-trinitrophenol (TNP), methylene blue (MB), Rhodamine B (RHB), and Methyl Orange (MO). We conducted the experiments at different timings, and machine learning algorithms, including Linear Regression (LR), Support Vector Machines (SVM), Gradient boosted machines (GBM), Random forest (RF), and XGBTree (XGB), were used to predict the catalytic activity. The performance of these algorithms was measured using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The results showed that the XGB algorithm performs best with NP and DNP. RF algorithm performs best with TNP, MB, and RHB, and the SVM algorithm performs best with MO. PdO-NiO bimetallic catalyst showed 98% reduction efficiency of azo compounds mixture within 8 min. Hence, we found PdO-NiO to be an efficient catalyst for real-site applications.