How to become a data scientist

1. 1Build a math foundation in high school

   Time: 4 years (high school)Sources 1

   Take the most advanced math your school offers, ideally through calculus, plus a statistics course if one is available. BLS points to calculus, linear algebra and probability and statistics as core preparation. Try programming early: free Python tutorials and small projects with public data, such as sports or weather records, teach you how messy real data is and whether you enjoy the work.

2. 2Earn a quantitative bachelor's degree

   Time: 4 yearsCost: $45,864 in median public tuitionSources 1, 2

   Common majors are statistics, mathematics, computer science, data science, economics and engineering. Whatever the major, take probability, statistical inference, linear algebra, programming in Python or R, databases and SQL, and at least one machine learning course. College Scorecard lists 197 bachelor's programs in data science specifically. A minor or electives in a field like finance, biology or public health can help, because some employers want industry knowledge.

3. 3Build a portfolio and get internship experience

   Time: During college and afterSources 1

   Hiring managers want proof you can work with real data. Publish two or three projects on GitHub that show the full process: getting and cleaning data, analysis, a model, and a clear write-up of what it means. Summer internships in analytics or data science are the most direct route to a full-time offer. Research assistant work with a professor also counts and helps if you plan on graduate school.

4. 4Start as an analyst or junior data scientist

   Time: 1-2 yearsSources 1

   Many people begin as data analysts, business intelligence analysts or junior data scientists. These jobs build strong SQL skills, knowledge of how a business works and experience presenting results, which matter as much as modeling. Ask to take on forecasting, experiments or predictive projects so you can show model-building experience when you apply for data scientist roles.

5. 5Consider a master's degree

   Time: 1-2 yearsCost: Varies; median federal debt for data science master's graduates is $38,994Sources 1, 2

   A graduate degree is expected, or at least favored, by some employers, especially for research, advanced machine learning and senior roles. Options include master's degrees in data science, statistics, computer science or analytics, and many are offered part time or online for working professionals. Compare curricula closely: look for real statistics and machine learning depth, not only software tools.

6. 6Add certifications if they fit your target employers

   Time: Weeks to a few monthsCost: $200-$695Sources 3, 4

   Certifications are optional but can help show cloud or applied skills. The Google Cloud Professional Machine Learning Engineer exam costs $200, runs two hours with 50 to 60 questions and is valid for two years. INFORMS's CAP-Expert exam costs $440 for members or $640 for nonmembers, plus a $55 application fee, and requires analytics work experience. Neither replaces a degree or portfolio.

Programs, tuition and graduates’ earnings

U.S. Department of Education, College Scorecard (most recent cohorts, released 2026-10-15). Scorecard groups programs by 4-digit field, so a row can include related programs. Tuition is the median published undergraduate tuition and fees of the colleges offering the program (shown only for undergraduate credentials); debt is the median of programs’ median federal loan debt at completion; earnings are the national median for the field and credential four years after completion.

Requirements differ by state. Check your state’s rules in the CareerOneStop License Finder (U.S. Department of Labor) before you enroll.

Statistics and probability

Knowing when a pattern is real, how to set up an experiment and how much to trust a model's output is what separates a data scientist from someone running code.

Python or R programming

Nearly all cleaning, modeling and visualization is done in code, and interviews usually include a coding exercise.

SQL

Most business data sits in relational databases and warehouses, so you need to pull and join it yourself before any analysis can start.

Machine learning

Classification, regression and forecasting models are core tools, along with the judgment to choose a simple model when it works just as well.

Explaining results to non-specialists

A model only matters if managers understand it well enough to change a decision, so short write-ups and clear charts are part of the job.

O*NET 31.0 rates these skills as most important for the occupation: mathematics, critical thinking, reading comprehension, speaking, active learning.

Training cost

$45.9K–$84.9K

Before financial aid

Time

5-6 years

Typical path

Median pay

$120,230

All US jobs: $50,980

Payback once qualified

0.7–1.2 yrs

Pay above the all-jobs median vs. cost

Low: 4 years at the Scorecard median public in-state tuition for data science bachelor's programs ($11,466 a year = $45,864). High: the same bachelor's plus the Scorecard median federal loan debt for data science master's graduates ($38,994) as a stand-in for master's costs. Not included: housing, private tuition, and optional certification fees (for example $200 for the Google Cloud ML Engineer exam).

Rough guide only: ignores taxes, aid, interest and earnings given up during training. How we calculate it.

• Senior or lead data scientist

  With several years of experience, data scientists lead larger projects, set modeling standards and mentor junior staff.

• Machine learning engineer$135,980 median

  Data scientists with strong software skills move into building and running production ML systems, which takes deeper engineering, testing and cloud infrastructure knowledge.

• Business analyst or analytics manager$101,860 median

  Those who enjoy the business side can move into management analysis and strategy roles, or manage analytics teams, with more meetings and less hands-on coding.

• Actuary$130,000 median

  Data scientists in insurance with strong probability skills can move into actuarial work, but it requires passing a series of professional actuarial exams.