This project aims to identify the key features contributing to production issues in silicon wafer manufacturing using machine learning. By leveraging Random Forests and XGBoost algorithms, we analyze various production parameters to highlight the most significant factors affecting the yield and quality of silicon wafers. Understanding these feature importances can lead to more informed decisions to mitigate production problems. Additionally, SMOTE (Synthetic Minority Over-sampling Technique) is employed to ensure a balanced dataset for more reliable analysis.
- Random Forest Analysis: Utilizes the Random Forest algorithm to evaluate feature importance based on multiple decision trees, providing insights into the stability and contribution of each feature towards production outcomes.
- XGBoost Analysis: Applies the XGBoost technique for a more sophisticated, gradient-boosting framework to assess feature significance and performance in predicting production issues.
- Feature Importance Evaluation: Compiles and compares insights from both models to identify the most critical features affecting silicon wafer production, offering a comprehensive understanding of the underlying causes of production inefficiencies.
Data and images within this repository have been intentionally censored. This censorship is necessary to protect sensitive company information and ensure compliance with data security policies.