Application of machine learning in predicting corrosion inhibition capacity of Spinacia oleracea leaf extract on copper

In many different industries, material corrosion is a major problem and it has a big financial impact. In this case, plant extracts employed as corrosion inhibitors, provide affordable way to prevent copper from corroding in acidic media, providing an alternative to the dangerous chemicals now in use. This study presents an integrated experimental and machine learning approach for investigating the corrosion inhibition performance of Spinacia oleracea leaf extract on copper in nitric acid medium. Experimental procedure involving gravimetric analysis under different concentrations, temperatures, and exposure durations, inhibitory efficiency was conducted. Different machine learning (ML) models, namely Artificial Neural Network (ANN), Random Forest (RF), Support Vector Machine (SVM), and Decision Tree (tree) were developed for predicting the corrosion rate. To overcome the black-box limitation of the ML models, an interpretable feature analysis was carried using Shapley Additive ExPlanations (SHAP).The accuracy and validity of the models were evaluated using statistical tests like Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Deviation (MAD), Mean Absolute Percentage Error (MAPE), and Variance Accounted For (VAF).The best prediction accuracy was obtained with SVM, giving an averaged validation-based RMSE, MAE, MAD, MAPE, and VAF values of 0.386, 0.287, 0.192,0.08773, and 99.093, respectively, across 5 folds. SHAP interpretability identified inhibitor concentration as the most influential variable controlling corrosion inhibition. The data-driven framework that combines experimental gravimetric analysis with SHAP-enhanced ML in this study contributes to the broader development of transparent, eco-friendly, and data-driven corrosion prediction models.