Fig. 8: Performance evaluation of the BN Model using stratified CV. | npj Materials Degradation

Fig. 8: Performance evaluation of the BN Model using stratified CV.

From: Predicting stress corrosion cracking in downhole environments: a Bayesian network approach for duplex stainless steels

Fig. 8

a Confusion matrix displaying the aggregated results from test sets regarding the pitting corrosion node. Here, the BN model predicted 1468 TP and 1173 cases TN, yielding 90.34% recall and 91.43% specificity. b ROC curves for each CV fold for the pitting corrosion node, where AUC scores range from 0.936 to 0.960, and an average AUC of 0.950 ± 0.008. c Confusion matrix displaying the aggregated results from test sets regarding the SCC node. The BN model predicted 1069 TP and 1598 TN, resulting in 90.21% recall and 92.74% specificity. d ROC curves for each CV fold for the SCC node, where AUC scores range from 0.931 to 0.958, and an average AUC of 0.945 ± 0.008. The overall accuracy for the pitting corrosion and SCC nodes was 90.21% and 91.39%, respectively.

Back to article page