15 Data Science Research Ideas for High School Students
Data science research ideas for high school students are easier to find than most people think, because almost every subject now produces data worth studying. A 2026 Mathematica study, conducted with Data Science for Everyone, found that 88% of parents, teachers, administrators, and education officials believe every student should develop data literacy before they graduate high school, and 76% say the rise of AI has made data education more important than it was. Only 58% think schools currently give students enough opportunity to build those skills. [1]
Scholar Launch mentor Dr. Ganesh Mani, who teaches at Carnegie Mellon University, shared that study with our team, and his recommendation is the same: every student should be data literate, whatever they plan to major in. The gap between what students need and what most classrooms offer is exactly the space an independent research project can fill.
Short answer: Data science is a method, not a major. Any field that produces records, surveys, images, or text can be studied with it, which means a student interested in psychology, history, or media has as much to work with as a student headed for computer science. Below are 15 research ideas grouped by field, each one small enough for a first serious project.
Why data literacy matters outside of STEM
The career argument is easy to make. The Bureau of Labor Statistics projects that employment of data scientists will grow 35% between 2025 and 2035, against 3% for all occupations combined. [2]
The intellectual argument matters more. Learning to work with data teaches you to ask a question precisely, to notice when your evidence cannot answer it, and to tell the difference between a pattern and a coincidence. Those habits travel. A historian who can analyze a century of census records, a journalist who can audit a public dataset, and a psychologist who can design a clean experiment are all doing the same underlying work.
You also do not need a large dataset or an expensive tool to begin. Most of the ideas below can be pursued with public data, a spreadsheet, and a willingness to be careful.
STEM: physical and life sciences
1. Air quality and traffic patterns in your own city
Public sensor networks publish hourly air quality readings, and many municipalities publish traffic counts. Compare the two across a year and you can ask whether a specific road closure, school schedule, or construction project measurably changed what people in a neighborhood were breathing.
2. Predicting crop yield from weather records
Agricultural yield data and historical weather data are both freely published by government agencies. Build a simple model that estimates yield from rainfall and temperature, then test where it fails. The failures are usually the interesting part of the project.
3. Genomic data and disease risk
Open biological databases let students explore how researchers connect genetic variants to disease risk. A realistic project scope is to replicate a published association on a public dataset and examine how sensitive the result is to the choices the original authors made.
Psychology: behavior and cognition
4. Sleep, screen time, and academic performance
Design a short survey for students at your school, collect responses anonymously, and look for relationships between sleep duration, evening screen use, and self-reported focus. The methodological question is the real one: what can a self-reported survey actually establish, and what can it not?
5. Decision-making under time pressure
Build a simple online task where participants answer questions with and without a countdown timer, then compare accuracy and confidence. Behavioral economics has decades of findings here that a student can test on a small scale.
6. What makes a habit stick
Track a single behavior across several weeks for a small group of volunteers and model the drop-off. Questions about streaks, reminders, and social accountability are all testable with modest data.
Social science: communities and policy
7. Teen social media use and wellbeing
Pew Research Center reported in April 2026 that about three-in-ten teen TikTok users say they spend too much time on the platform. [3] A student project can dig underneath a national number like that by surveying a local population and asking which specific uses, not which platforms, correlate with feeling worse.
8. Housing costs and school enrollment
Local housing price data and district enrollment records are usually public. Studying them together lets you ask how quickly neighborhood change shows up in classrooms, and which schools absorb the most of it.
9. Auditing a public transit system for equity
Transit agencies publish schedules and route data in a standard format. Map how long it takes to reach a hospital or a grocery store from different neighborhoods and you have an accessibility analysis that mirrors how urban planners actually work.
Business and media: markets, platforms, and audiences
10. What moves a stock beyond the fundamentals
Compare a company's share price against news volume, product launches, or social sentiment over a defined period. The useful finding is often the absence of a relationship, which teaches a lesson about markets that no textbook delivers as well.
11. Portfolio construction and risk
Using historical returns, build several portfolios with different risk profiles and test how they would have performed through a specific downturn. The mathematics is accessible, and the results are genuinely surprising to most first-time analysts.
12. How recommendation systems shape what you see
Log what a platform serves you over two weeks under deliberately varied behavior, then analyze how quickly the feed narrows. This is a study of design, not just of algorithms, and it produces data nobody else has.
Humanities: text, history, and culture
13. A century of newspaper language
Digitized newspaper archives let you measure how coverage of a single topic changed in volume and vocabulary across decades. Text analysis turns a question historians debate qualitatively into one you can chart.
14. Authorship and style
Computational stylometry uses word frequency and sentence structure to attribute anonymous or disputed texts. Students can test the method on works of known authorship first, which is the honest way to learn whether to trust it.
15. Mapping a historical migration
Census and immigration records support projects that trace how a community moved and settled over generations. Combining that with maps produces work that is both analytical and genuinely readable.
Choosing the right idea for you
Notice what these 15 have in common: none requires a laboratory, and none requires you to already call yourself a data scientist. What they require is a question you actually want answered and the patience to test it properly.
The most common mistake students make is starting too broad. "How does social media affect teenagers" is a topic, not a research question. "Do teenagers who use social media primarily to message friends report different wellbeing than those who primarily scroll" is a question you can answer in a semester. Narrowing well is the hardest step, and it is the first thing a mentor should work on with you.
Ready to start?
Scholar Launch pairs students with university faculty who help turn a broad interest into a focused project. If one of the ideas above caught your attention, our upcoming programs run in adjacent territory: Applied Mathematics in Finance covers portfolio optimization and modeling, Platform Strategy, UX Design, and Digital Innovation examines how digital systems shape behavior, and Fundamental AI and Machine Learning builds the Python and modeling foundation the rest of this list depends on.
If you would rather build something entirely your own, explore our Custom 1-on-1 Research Program or apply now to get started.
Frequently asked questions
Do I need to know how to code to do a data science research project?
No. Several ideas above can be completed in a spreadsheet, and most students who do learn to code pick up enough Python or R during the project itself. Curiosity about the question matters more than prior technical background.
Where can high school students find datasets they are allowed to use?
Government agencies, research universities, and open data portals publish enormous amounts of material at no cost. Census data, public health records, transit feeds, and digitized archives are all open, and a mentor can point you toward the ones that fit your question.
How long does a data science research project take?
A focused project generally takes about 12 weeks from question to finished paper. Projects that stall almost always stalled at the start, because the question was too broad to answer with the time available.
Is a data science project useful if I am not planning to study data science?
Yes. Admissions readers respond to evidence that a student can carry an original question through to a defensible conclusion. The subject can be history or psychology; the method is what demonstrates rigor.
References
[1] Deacon, G., Conroy, K., & Weissman, H. (2026, August 10). What students need to thrive in a data- and AI-driven world. Mathematica. https://www.mathematica.org/publications/what-students-need-to-thrive-in-a-data-and-ai-driven-world
[2] U.S. Bureau of Labor Statistics. (2026). Data scientists: Occupational outlook handbook. https://www.bls.gov/ooh/math/data-scientists.htm
[3] Pew Research Center. (2026, April 15). Teens' experiences on TikTok, Instagram and Snapchat. https://www.pewresearch.org/internet/2026/04/15/teens-experiences-on-tiktok-instagram-and-snapchat/
