Featured Research & Projects

Health Data Experimentation

R | Full Factorial DOE | ANOVA | Linear Modeling

· Designed and conducted a \(2^4\) full factorial study over 16 nights on temperature, screen exposure, bedtime, and light conditions on sleep quality; implemented half-normal plots and two-step optimization to identify.

· Estimated main and interaction effects with ANOVA & OLS model, implementing regression refitting and applying nominal-the-best optimization to balance mean sleep score and variance stability.

· Achieved \(R^2\) of 0.81 with 5-fold cross-validation and all retained key factors significant (\(p < 0.05\)).

LLM-Based Surgical Scheduling Optimization

Python | CPLEX | Zhipu API | COE

· Developed an API-based scheduling framework integrating GLM-4 model with operations research to automate surgical case analysis, surgery sequencing and medical machine allocation, handling requests in real time.

· Applied the Chain of Expert multi-agent framework including constraint engineer, data validator and mixed-integer programming to translate policies and surgeon information into constraint templates and model selection.

· Linked LLM outputs to MILP in CPLEX, tested with 1k+ surgeries across 20 rooms and improved scheduling efficiency by 18%, resource utilization by 22%, and real-time decision speed by 20%.

Beverage Chain Store Database

Python | Tableau | SQL

· Implemented an comprehensive database (ER diagram, schema, data dictionary) for a beverage chain store and 30+ SQL queries to automate core workflows for customers, staff, suppliers and 3rd party logistic providers.

· Built Tableau dashboards and ran OLS regression to estimate order price, identified main customers to be 14-29 years old and statistically significant price factors to be gender (+20.96 rmb) and offline ordering (+17.25 rmb).

NUBS Strategy Consulting Program

Nottingham Advantage Award

Python | NLP

· Led an 8-member team and built a Python KPI model tracking efficiency (throughput/hour, on-time completion, rework rate) and inclusion (retention, promotion rate, training-completion gap) of Inclusion Factory’s workforce.

· Quantified performance drivers with multiple regression and K-means clustering and implemented text analytics on employee feedback using TF-IDF and logistic regression and sentiment analysis using BERT model.

· Authored a 16,000+ words data-driven consulting report, integrating quantitative and textual insights, recognized under the Nottingham Advantage Award and adopted for the client’s strategic planning.