Biography

I am a undergraduate researcher at University of Connecticut. My research interests lie primarly in city- and building-scale disaster prediction. I an anticipated to received my B.A. from the Statistics Department and Economics Department. University of Connecticut.

Research Experience

05/2024 - Present | Research Assistant | Department of Statistics, University of Connecticut

• Conducted preliminary research for city-scale and building-scale disaster prediction by estimating building heights, reprocessing the UT-GLOBUS dataset, developing interactive maps, and performing exploratory spatial data analysis.
• Preprocessed satellite data with QGIS and SNAP, combined with other datasets including OpenStreetMap building footprints, and applied Random Forest Regression to predict urban building heights for disaster modeling.
• Reprocessed data of 413 major U.S. cities (gpkg files) from UT-GLOBUS datasets using HPC, calculating coordinates for each building and exporting attributes of each city into individual CSV files.
• Developed an interactive map using data from New Haven to visualize building footprints and associated attributes. Integrated exploratory spatial data analysis across 4 cities to examine spatial autocorrelation and enhance visualization insights.
• Created a detailed tutorial to ensure reproducibility of data preprocessing, integration, and analysis workflows.

Internship Experience

01/2024 - 05/2024 | Intern | Qian Shi Du

• Performed time series analysis to compare current and historical trends and developed a GPT-based model to forecast customer demand patterns, with SHAP values used to interpret feature importance.
• Employed K-means to segment customer preferences, guiding targeted product development and marketing.
• Integrated insights from customer segmentation, purchasing behavior, and product tracking to forecast sales trends, supporting the company’s strategic decision-making in international trade.

05/2024 - 08/2024 | Intern | GBCS-SkyIT

• Developed backend services for the Voop application using Express.js and Django REST, contributing to scalable architecture and API integration.
• Collaborated with teammates in migrating database from Firebase to MySQL, optimizing database schema, and ensuring data accuracy during the transition.
• Built and tested APIs for Voop, utilizing Postman to debug and validate functionality.

Project

Bytedance - Data Optimization and Algorithm Enhancement | Remote Collaboration

• Preprocessed large-scale text data, using techniques such as tokenization and stemming to improve data quality and enhance search engine performance.
• Applied a SAITS model for data imputation and fine-tuned a LLaMA2 model by LoRA to predict product prices based on descriptions, addressing missing data challenges.
• Optimized product price prediction algorithms by refining code efficiency and improving data pipelines, resulting in increased accuracy and operational performance.

Exploratory Data Analysis in NYC Open Data—Rodent Inspection | Statstics Course (Introduction to Data Science)

• Cleaned and analyzed NYC rodent inspection data using methods like binomial regression models.
• Designed an interactive map with GeoPandas to display rodent sighting locations with all related information and displayed the result to the NYC Open Data team to support rodent inspection.

Prediction of NBA Players’ Salary based on On-Court Contributions | Statstics Course (Introduction to Statistical Learning)

• Cleaned and conducted exploratory data analysis on on-court statistics to examine their correlation with player salaries.
• Utilized 4 machine learning models and applied methods such as hyperparameter tuning, Bayesian optimization, and model ensembling to improve prediction accuracy.

Miscellaneous

Activities I find enjoyable: baskerball, reading, writing, and more.