@article{ART003397468},
author={DongGug Yoo},
title={Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2026},
volume={31},
number={10},
pages={281-288}
TY - JOUR
AU - DongGug Yoo
TI - Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes
JO - Journal of The Korea Society of Computer and Information
PY - 2026
VL - 31
IS - 10
PB - The Korean Society Of Computer And Information
SP - 281
EP - 288
SN - 1598-849X
AB - This study re-evaluated five classifiers for crack occurrence in the supplied laser-cladding dataset (1,000 records; 500 crack-positive and 500 crack-negative cases). A 12-predictor model including post-process coating properties was compared with a six-predictor model restricted to candidate pre-process inputs. The same stratified 700/300 training/test split and five training folds were used across all analyses. Imputation, normalization and dummy encoding were estimated within each training fold. Model selection used cross-validated ROC AUC, and a common probability threshold of 0.5 was applied. The cross-validation-selected pre-process SVM achieved a test AUC of 0.848 (95% CI 0.806– 0.890), accuracy of 0.740, and F1-score of 0.738. The 12-predictor unpenalized logistic model achieved an AUC of 0.996 and accuracy of 0.983, but exhibited complete separation in the training data and nonconvergence in the original fitting configuration. The performance gap demonstrates the importance of predictor availability at the intended prediction time; it does not establish causal effects or industrial deployment validity. Dataset provenance, label acquisition, sample independence and external validation remain necessary before operational conclusions can be drawn.
KW - Laser cladding;crack classification;predictor timing;data leakage;cross-validation
DO -
UR -
ER -
DongGug Yoo. (2026). Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes. Journal of The Korea Society of Computer and Information, 31(10), 281-288.
DongGug Yoo. 2026, "Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes", Journal of The Korea Society of Computer and Information, vol.31, no.10 pp.281-288.
DongGug Yoo "Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes" Journal of The Korea Society of Computer and Information 31.10 pp.281-288 (2026) : 281.
DongGug Yoo. Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes. 2026; 31(10), 281-288.
DongGug Yoo. "Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes" Journal of The Korea Society of Computer and Information 31, no.10 (2026) : 281-288.
DongGug Yoo. Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes. Journal of The Korea Society of Computer and Information, 31(10), 281-288.
DongGug Yoo. Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes. Journal of The Korea Society of Computer and Information. 2026; 31(10) 281-288.
DongGug Yoo. Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes. 2026; 31(10), 281-288.
DongGug Yoo. "Performance Evaluation of an AI Classification Model for Predicting Laser Cladding Process Outcomes" Journal of The Korea Society of Computer and Information 31, no.10 (2026) : 281-288.
