<![CDATA[Corrosion is one of the main problems in various industries, causing increased production costs, maintenance expenses, and decreased equipment efficiency. Inhibitors, especially organic compounds, have become an increasingly sought-after solution to reduce corrosion in an effective and environmentally friendly manner. This research aims to compare linear algorithms based on multilinear regression (MLR) and nonlinear algorithms based on decision tree regressor (DTR) for a case study of predicting corrosion inhibition efficiency (CIE) values of pyridazine derivative compounds as corrosion inhibitors. This research is a data-based theoretical study using a machine learning (ML) approach. In this study, we used a dataset of pyridazine compounds from published literature consisting of 20 pyridazine compounds with molecular properties as features (independent variables) and CIE values as targets (dependent variables). The analysis consists of analyzing the prediction model's performance and important features that support model performance. The research results show that the DTR-based nonlinear model has better performance than the MLR-based linear model based on evaluation metrics and prediction data distribution plots. We also find that the molecular properties fraction of electron transferred (∆N) and electron affinity (A) respectively emerge as the features most responsible for the prediction performance of the DTR model. We conclude that the DTR-based nonlinear algorithm can be used as an accurate and reliable predictive model to predict the corrosion inhibition ability for potential new pyridazine derivative compounds.]]>
