Advanced and explainable machine learning model for prediction of surface roughness of tempered steel AISI 1060

This research examined the performance of a machine learning algorithm when predicting the surface roughness of tempered steel AISI 1060. Different machine learning algorithms, such as decision tree (DT), random forest (RF), adaptive boosting (ADB), gradient boosting (GB), and extreme gradient boosting (XGB), were optimized by using 10-fold cross-validation and the grid search method. From these optimized models, the decision tree, adaptive boosting, gradient boosting, and extreme gradient boosting were used as base models to develop a more powerful machine learning model called super learner machine learning. The linear regression (LR) was used as a meta-model in developing super learner machine learning. The developed super learner model performance was then validated against all machine learning models used in this research. For performance measurement metrics such as mean absolute error (MAE), root mean square error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R²) has been used. The developed super learner model achieved the highest R² of 99.2% and the lowest MAPE of 2.6% on the test data set when compared with other machine learning models. Further, the SHAP method shows that hardness has the highest effect, followed by feed rate and cutting speed, respectively. Most machine learning approaches are not used practically for user applications, but in this research, a graphic user interface framework called fast, accurate, and intelligent (FAI) frame was developed to predict the surface roughness of tempered steel AISI 1060. This research is used for practical application for any user in industry and for research purposes.