Cynthia Ng

cynthia ng

With a background in international relations and data science, I like connecting the dots to deepen our understanding of ourselves and the world.

My dream is to tell stories about our shared humanity.

recent work

• Standardized vaccine safety assessment process for FDA by creating 3 scripts in R, saving 240 hours annually.
• Documented and presented methodology to healthcare consultants, influencing strategy of 85+ analysts.
•   Launched 3 plant-based meat products in 650 stores, achieving 3rd highest sales among new competitors.
•   Analyzed sales data and identified low-performing stores, driving a 58% monthly revenue increase.
•   Forecasted sales using self-built template for global operations planning; achieved 88% accuracy.
•   Managed 3 trade show exhibits: secured $218k budget, collaborated with partners, and gathered 755+ leads.
• Led delivery-only kitchen expansion strategy: identified 50 restaurant partners and 7 expansion areas using Looker, resulting in 100% increase in kitchen sites.
• Improved restaurant performance by analyzing operational data of 150+ restaurants and providing tailored recommendations like menu optimization, increasing sales for 35 brands.
• Developed 40-page recommendations to charity’s child safety program based on research and interviews.
• Created 40 bilingual child protection resources, benefiting 80+ families in inaugural program.
• Tracked pandemic response of South Korea and Hong Kong, culminating in 15-page memo and 2-page summary.
• Developed teaching case for Prof. J. Nicholas Ziegler’s graduate MPA course.
• Communicated with 540+ constituents: drafted letters, wrote legislative summaries, and managed front desk.
• Wrote 100+ memos for 93+ committee hearings and meetings with stakeholders; memos included stakeholder research and background information to prepare Congressman David N. Cicilline on special interest topics.

data projects

Taylor Talk

As Swifties refreshed Ticketmaster for Eras Tour seats, my team combined our love for Taylor Swift’s music and community with our AWS cloud computing skills. We dived into features of Taylor Swift’s music and posts in her Reddit community, building models that flagged controversial posts and discrimination. Our natural language processing analysis also revealed how different moods and reactions flowed through the community, especially during tour announcements and album releases.
Language: Python
Tools: AWS, Amazon SageMaker, PySpark, scikit-learn, NLTK, RegEx, Pandas, and numpy

Project Website
Walkability in Washington, DC

We explored how walkability correlates with people's lives in Washington, D.C. By analysing walkability data and conversations on Reddit about transportation, we discovered areas where people could easily walk and bike tended to have healthier, more prosperous outcomes. This highlighted a concerning pattern - many lower-income neighbourhoods lacked safe, walkable streets and bike lanes, which connected to poorer health in these communities. I shared these findings at the 2023 Evolution of Data Science Conference.
Languages: R and Python
Tools: Altair, Plotly, ggplot2, matplotlib, Seaborn, Pandas, numpy, geopandas, and SciPy

Project Website
Workforce Diversity

In light of massive layoffs, I was curious about what makes people stay or leave their jobs and obtained a large dataset on six public companies’ workforce from Revelio Labs. Through experimenting with various machine learning approaches - from straightforward Naive Bayes to more complex decision trees that achieved 92.8% accuracy - I uncovered interconnections between diversity, pay, and job satisfaction. One of my most intriguing findings challenged conventional wisdom: higher-paid employees were more likely to seek new opportunities elsewhere.
Languages: R and Python
Tools: Pandas, numpy, and scikit-learn

Project Website
Assessing Food Additive Safety with FDA

For my graduate Capstone project, I worked with the Office of Food Additive Safety within the Food and Drug Administration (FDA) to assess food safety. I used Natural Language Processing and Machine Learning techniques to analyse 30,000 scientific papers, identifying and extracting patterns between food products, pathogens, and adverse events. The FDA's Office of Food Additive Safety incorporated our findings into their work to spot potential food safety issues before they could become problems.
Language: Python
Tools: Pandas, NTLK, spaCy, and texacy

Project