This repository is an index of my Machine Learning coursework and portfolio projects at American International University-Bangladesh (AIUB). Each project is maintained in its own independent repository with its implementation, data, documentation, and reproducible outputs.
An academic group project comparing dengue-outcome prediction across three biomarker configurations—base demographic and residential features combined separately with NS1, IgG, and IgM. The project evaluates classical and ensemble classifiers, robustness checks, class-imbalance strategies, and model explainability.
Role: First Author, Team Leader, and Corresponding Student Author in a group project
Focus: Three-case biomarker comparison, demographic/residential classification, ensemble learning, evaluation, cross-validation, imbalance handling, and SHAP explainability
Technology: Python, pandas, scikit-learn, imbalanced-learn, SHAP, Matplotlib, and Jupyter
Key methods
- K-Nearest Neighbors and Logistic Regression
- Decision Tree and Random Forest
- Bagging, AdaBoost, and Gradient Boosting
- Reproducible 80/20 stratified train/test split
- Five-fold stratified cross-validation
- Random oversampling, random undersampling, and SMOTE-NC
- Accuracy, precision, recall, F1-score, MAE, confusion matrices, and classification reports
- Random Forest interpretation with SHAP
- Supervised classification and ensemble learning
- Data preprocessing with leakage-aware pipelines
- Model evaluation and cross-validation
- Class-imbalance analysis
- Model explainability and data visualization
- Reproducible academic machine-learning workflows
This repository contains links and concise summaries only. Project code, datasets, notebooks, papers, and results remain in their independent repositories; they are not duplicated here or included as submodules.