Abstract
Hydrogen sulfide (H2S) can be extremely flammable and hazardous. It is frequently found in industrial settings where individuals face serious dangers to human health and safety, such as in the production of oil and gas, wastewater treatment, and chemical manufacture. H2S can create serious health risks, even at low quantities, such as respiratory failure and fatalities, which makes gas detection imperative. In this study, ML models have been employed to predict how n-type materials doped with various metals—such as Ag, Pd, and Au—will react when they are exposed to hydrogen sulfide (H2S) gas. The tree-based machine learning models used in the study include Gradient Boosting Regression (GBR), Random Forest (RF), and Decision Tree (DT) coupled with several optimization strategies (such as Bayesian optimization, Random Search, and Grid Search). Additionally, the examination involves the utilization of Partial Dependence Plots (PDPs) and Shapley Additive exPlanations (SHAP) to gain insights into the intrinsic mechanisms of the ML model and to elucidate the opaque nature of the ML model, these post-model interpretations elucidate the significance of features in gas sensing.