Grace Jihyun Lee

Data analyst · May to Jun 2026

Campaign analysis for StratusLink Communications during the COOP fellowship: 728 reports across 13 variables covering creative sizes, inventory sources, audiences, and states.

• The 320x50 creative had the lowest CPA of any format at $311, against a campaign average of $327, and performed best on mobile.
• Ranked exchanges and publishers with a weighted KPI score and recommended direct buys with the top five of each, plus more budget in Indiana, Kentucky, and Ohio, where CPAs were lowest.
• Built the analysis in Excel PivotTables and 10 Tableau visualizations and presented it to the client with two other analysts.

• Excel
• PivotTables
• Tableau

Data analyst · Apr to May 2026

Team analysis of the 2018 San Francisco crime dataset, more than 150,000 reports across 27 variables, to find what, where, and when crime happens most.

• Larceny theft was 38.5% of all incidents, peaking in July and August, on Fridays, and between 12 to 1 PM and 6 to 7 PM.
• The Mission neighborhood had the most incidents (13,916), and 84% of all cases were still open. Recommended more patrol and community programs in the northern neighborhoods.
• Cleaned duplicates and misspellings in Excel and built 16 Tableau visualizations, including a neighborhood map and hourly and monthly time series.

• Excel
• Tableau

Researcher · Mar to Jun 2024

An 11-week research project testing whether urbanization drives economic growth, with a six-variable OLS model of GDP per capita across countries.

• Compiled a 195-country, 13-variable dataset in Excel from more than 10 sources, then cleaned it in Python to a 157-country sample.
• Urbanization's coefficient was positive and significant at the 5% level, and the model explained 42% of the variance in GDP per capita.
• Wrote up expected signs, results, and limits, including an unexpected negative coefficient on education.

• Python
• Excel
• OLS regression