This repository contains all the materials for a machine learning project focused on the implementation and optimization of Support Vector Machines (SVMs) and Bagging-SVM for classification tasks. The project, entirely developed by Courtney H., explores advanced techniques to improve model performance across multiple datasets.
Repository Contents
- Machine Learning Project #2 (File: .pdf)
- A PDF version of a PowerPoint presentation that details the project's methodology, implementation, and results.
- Covers the process of applying SVM and ensemble Bagging-SVM techniques to datasets such as credit risk and cybersecurity data, including a comprehensive analysis of performance metrics like F1-micro, F1-macro, and balanced accuracy.
- Machine Learning Project (File: .ipynb)
- A Jupyter Notebook containing the complete code for the project.
- Includes markdown explanations throughout the notebook for clarity and reproducibility, walking through data preprocessing, model training, evaluation, and performance comparisons.
- Original Work: This project was completed solely by Courtney H., with no external contributions.
- Techniques Applied: Explores SVM, Bagging-SVM, and voting-based ensemble methods to optimize classification performance.
- Metrics Achieved: Demonstrated 94% balanced accuracy for credit risk data and significant improvements in F1 scores using ensemble techniques.