作品集 · Selected work

Projects in bloom

A selection of projects spanning data analytics, machine learning, and applied AI. Each one turns raw data into clear, actionable decisions.

不正↗

Healthcare Provider Fraud Risk Explorer

End-to-end ML workflow that turns real Medicare Part B billing data into a ranked, explainable worklist of providers most likely committing fraud. Built a 6M provider-year panel (CMS + OIG LEIE) with peer-relative features, then trained Logistic Regression, Gradient Boosting, and XGBoost — optimizing precision-at-top-k for extreme class imbalance (~0.02% fraud).

Python
XGBoost
scikit-learn
Feature Engineering
Imbalanced Classification

融資↗

Bank Loan Risk & Performance Dashboard

Built an interactive Excel dashboard monitoring a retail bank's 38.6K-loan portfolio ($435.8M funded). Cleaned and modeled the loan book with Pivot Tables and KPI cards to expose a 13.82% bad-loan rate, isolate charge-off risk by purpose and term, and translate the findings into concrete underwriting recommendations.

Excel
Pivot Tables
Data Cleaning
Dashboard Design
Risk Analysis

株価↗

Monte Carlo Stock Price Simulator

Forecast short-term stock prices as a probability distribution instead of a single number. Estimated volatility from daily log returns of live Yahoo Finance data, then simulated 100,000 independent price paths forward to derive an expected price and a 95% confidence interval. Backtested on GOOG: the two-day forecast of $335.78 landed within two cents of the actual $335.76.

Python
NumPy
Monte Carlo
yfinance
Probabilistic Forecasting

画像↗

CNN-Based Medical Image Classification

End-to-end medical image classification pipeline in PyTorch that labels chest X-rays as Normal or Pneumonia. Built a custom CNN over the 28×28 grayscale PneumoniaMNIST (MedMNIST v2) dataset, covering preprocessing, training, evaluation, and inference on unseen test images — with the trained weights saved for reuse.

Python
PyTorch
CNN
Deep Learning
Medical Imaging