Fig 1.
Concept of the Loss function in the case of Supervised Machine Learning and Physics Machine Learning approach.
Fig 2.
Schematic representation of the Additive Friction Stir Deposition process.
Fig 3.
The machine learning framework implemented in the present work.
Fig 4.
Obtained heat map for regression-based algorithms.
Fig 5.
Feature importance plot obtained for regression-based algorithms.
Fig 6.
Decision Tree plot obtained in the present work.
Fig 7.
Actual vs Predicted value plots for a) Support Vector Regression, b) Decision Tree Regression, c) Random Forest Regression, d) XGBoost Regression, e) CatBoost Regression, f) AdaBoost Regression, g) Extra Tree Regression and h) Gradient Boosting Regression algorithms.
Fig 8.
Residual plots for a) Support Vector Regression, b) Decision Tree Regression, c) Random Forest Regression, d) XGBoost Regression, e) CatBoost Regression, f) AdaBoost Regression, g) Extra Tree Regression and h) Gradient Boosting Regression algorithms.
Fig 9.
Q-Q plots for a) Support Vector Regression, b) Decision Tree Regression, c) Random Forest Regression, d) XGBoost Regression, e) CatBoost Regression, f) AdaBoost Regression, g) Extra Tree Regression and h) Gradient Boosting Regression algorithms.
Table 1.
Results obtained from the supervised machine learning regression-based algorithms in the present work.
Fig 10.
Surface plots of the used Physics machine learning models in the present work.
Fig 11.
Contour Plot of Physics machine learning models used in the present work.
Table 2.
Results obtained from Physics based machine learning models.
Fig 12.
A correlation heatmap was obtained to classify the deposition quality in the present work.
Fig 13.
Feature importance plot obtained in the present work.
Fig 14.
Decision tree plot obtained for classification-based model.
Table 3.
Results obtained from the supervised classification-based machine learning algorithms to predict deposition quality.
Fig 15.
Confusion matrix plots obtained for a) Logistic, b) K-Nearest Neighbours, c) Support Vector, d) Stochastic Gradient Descent, e) Decision Tree, f) Random Forest, g) AdaBoost, h) Gradient Boosting and i) Stochastic Gradient Boosting algorithms.
Fig 16.
ROC plots were obtained for a) Logistic, b) K-Nearest Neighbours, c) Support Vector, d) Stochastic Gradient Descent, e) Decision Tree, f) Random Forest, g) AdaBoost, h) Gradient Boosting and i) Stochastic Gradient Boosting algorithms.