Elena (Yuxin) Liang

Designed, developed, and launched a webpage to showcase postdoctoral fellows’ research projects. Built the site architecture and layout using HTML, CSS, and JavaScript, incorporating responsive design and interactive features to ensure accessibility across devices. Utilized ngrok for seamless local development and testing.

Applied causal inference models (DiD, IV, PSM, RCT) using R to evaluate international development programs funded by USAID, performed statistical analysis (e.g., regression, hypothesis testing, causal inference). Developed comprehensive analysis reports and produced well-designed maps, figures, and tables.

Built and maintained end-to-end data pipelines integrating civic datasets (311 service data, GIS layers, AirSage mobility data) using Python (sklearn, PySpark) to predict illegal dumping incidents in Philadelphia based on 10 years of historical data. Built dashboards using plotly, seaborn, and matplotlib.

Supervised a student project leveraging the Gemini LLM API to classify trash categories, using PyTorch and TensorFlow for model fine-tuning via prompt engineering and hyperparameter optimization. Evaluated performance using precision, recall, F1-score, and confusion matrix analysis.

Executed ETL and managed data pipelines for ad hoc and diagnostic analysis using Stata and R (dplyr, lubridate), designing and analyzing large-scale A/B experiments, cohort analysis, and regression modeling to support policy evaluation. Drafted reports with R Markdown.

Designed quantitative survey and collected data using Qualtrics; conducted in-depth interviews and focus groups for qualitative analysis. Ensured high-quality data monitoring for metric movement.

Led A/B testing, user surveys, focus groups, and interviews to conduct quantitative and qualitative analysis.

Collaborated with cross-functional teams to analyze market trends across real estate, beverage, and electronics industries, translating findings into client reports, whitepapers, and presentations to guide strategic decisions.

Web scraped the IBDB using BeautifulSoup, compiling a dataset of 10,000+ shows and 130,000+ theater professionals, capturing details on show networks, tour stops, and artist relationships to analyze industry trends.

Designed ArcGIS StoryMap layouts and interactive maps, integrating Google Maps API to geocode theater locations. Created a visually engaging web-based geographic analysis showcasing Broadway’s long-term expansion and industry patterns.

Analyzed the impact of British colonial legacy on gender attitudes using USAID’s Demographic and Health Survey. Applied a two-stage least squares regression model with an instrumental variable approach.

Conducted data cleaning in Stata, and modeling in R using dplyr, lfe, spdep, splm, and ggplot2 for visualization.

Analyzed the evolution of social gender concepts using HistWords word embeddings and applied NLP techniques such as cosine similarity and vector arithmetic. Utilized pandas, numpy, sklearn, scipy.spatial, and matplotlib.