Fang Han

Independently reproduced the STABL (Sparse and Reliable Biomarker Discovery) framework to solve feature selection instability in high-dimensional omics data (p≫n).

Key Contributions:
Built a comprehensive Stability Selection pipeline in Python/Scikit-learn, leveraging sub-sampling techniques to approximate expected model output and ensure selection consistency.
Applied Lasso Regularization and Coordinate Descent to optimize the Bias-Variance Tradeoff by fine-tuning the hyperparameter λ, successfully reducing feature selection variance.
Implemented Synthetic Noise Injection mechanisms to establish rigorous signal-to-noise thresholds, effectively controlling the False Discovery Rate (FDR) in sparse environments.

Tech Stack: Python (NumPy, Pandas, Scikit-learn), Statistical Learning Theory, Regression Analysis.

Developed a Java-based application to streamline academic planning for UBC students, featuring course management, prerequisite tracking, and admission probability forecasting.

Engineered a robust persistence layer using JSON, enabling seamless saving and loading of complex, nested data structures (Student Profiles, Course Records, and Specialization Data).

Adhered to a rigorous Test-Driven Development (TDD) workflow, achieving 100% code coveragefor model and persistence layers using JUnit 5 and Jacoco.

Applied Object-Oriented Programming (OOP) principles to ensure a clean separation of concerns between the User Interface, Persistence, and Data Model componen

• Analyzed user behavior to identify choice overload patterns.
• Used MATLAB to build up a new customer model to predict people's choices under the new mathematical function designed.
• https://doi.org/10.1007/978-981-97-0523-8_39