Predicting elastic and electronic properties of quaternary Heusler alloys using machine learning and DFT calculations

Abstract The Decision Tree (DT) Regression model is employed to predict the elastic and electronic characteristics of quaternary Heusler alloys using pre-existing and newly developed databases. The previously discussed were used to gather, preprocess, clean, and prepare the data so that the model would be more productive and efficient. The performance of the (DT) model demonstrates its robustness against overfitting, outliers, multicollinearity, and distribution of data points, in addition to being accurate enough to predict the response variables on the test data. The parity plots comparing the machine learning predicted values with the computed values using density functional theory (DFT) exhibit a linear relationship, with adjusted R 2 and MSE values predominantly lying within the range of 0.80 to 0.93 and 0.10 to 0.22 for all predicted properties of the quaternary Heusler alloy. In light of the results anticipated from the model analysis, a Validation is cited comparing ML predicted outcomes with first principle calculations based on (DFT) by selecting MnFeNiAl QHA, indicating the DT model’s excellent precision with a 13% average error, which suggests high accuracy of results predicted by the model.