Analysis and Prediction of Product Quality Using Statistical Analysis and Machine Learning with Data Supplied by the Quality Control Unit of Mobarakeh Steel Company, Isfahan

Abstract

The present study aims to analyze and predict the quality of steel products through the simultaneous application of statistical analysis methods and machine learning algorithms, utilizing real operational data obtained from the Quality Control Unit of Mobarakeh Steel Company in Isfahan. The objective is to enable early defect prediction and improve production processes. In this research, production data—including process variables, raw material characteristics, operating conditions of production lines, and quality control test results—were collected. Following data cleaning and preprocessing, the dataset was analyzed using descriptive and inferential statistical methods. Subsequently, machine learning models such as regression algorithms, decision trees, and artificial neural networks were employed to predict product quality. The findings indicated that integrating statistical analysis with machine learning models effectively identifies patterns influencing product quality and enhances prediction accuracy regarding product quality status. As a result, this approach contributes to reducing production waste, improving process control, and increasing production line productivity. Ultimately, the proposed model can serve as a decision-support tool for quality control units and production management in steel industries and can facilitate the transition toward smart manufacturing and predictive quality control.