Answers are grounded in this paper and cite their sources.
Purpose This study proposes a machine learning framework to predict customer complaints from production line tests in an automotive company's lot-release process, enhancing Quality 4.0. It aims to design and implement the framework, compare different machine learning (ML) models and evaluate a non-sampling threshold-moving approach for adjusting prediction capabilities based on product requirements. Design/methodology/approach This study applies the Cross-Industry Standard Process for Data Mining (CRISP-DM) and four ML models to predict customer complaints from automotive production tests. It employs cost-sensitive and threshold-moving techniques to address data imbalance, with the F1-Score and Matthews correlation coefficient assessing model performance. Findings The framework effectively predicts customer complaint-related tests. XGBoost outperformed the other models with an F1-Score of 72.4% and a Matthews correlation coefficient of 75%. It improves the lot-release process and cost efficiency over heuristic methods. Practical implications The framework has been tested on real-world data and shows promising results in improving lot-release decisions and reducing complaints and costs. It enables companies to adjust predictive models by changing only the threshold, eliminating the need for retraining. Originality/value To the best of our knowledge, there is limited literature on using ML to predict customer complaints for the lot-release process in an automotive company. Our proposed framework integrates ML with a non-sampling approach, demonstrating its effectiveness in predicting complaints and reducing costs, fostering Quality 4.0.
No takes yet. Share an insight, caveat, or question.
Lobo et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: