How to Become a Data Scientist

BLS says data scientists use analytical tools and techniques to extract meaningful insights from data. A typical project starts with deciding which data are available and useful, then collecting, categorizing, and analyzing them; creating, validating, testing, and updating algorithms and models; presenting findings with data visualization software; and making business recommendations to stakeholders.^[3]

Much of the early work is unglamorous. BLS notes that data scientists may start with large, unstructured datasets, commonly called raw data, and must clean them, structuring them so software can read them. Data can come from surveys, other organizations' databases, or web-scraping tools.^[3] The models come after that: BLS describes data scientists developing algorithms and models to support machine learning, using machine learning to classify data or make predictions, and testing the models for accuracy, including when new data arrives.^[3]

The last step is communication. BLS says visualization techniques let data scientists present their analyses to technical and nontechnical audiences, including colleagues, managers, and clients, and that ensuring the audience understands the information is what lets them recommend business decisions or process changes.^[3] A model nobody understands does not change a decision.

BLS also separates three specialties. Data scientists with a strong coding or engineering background may build machine learning systems and devise ways to enhance web-browsing functions; others conduct research for reports or academic journals; and others focus on business strategy for marketing, sales, and user engagement.^[3] The title is the same, and the day is not, so ask which one a job is before applying.

O*NET's work-context survey describes a desk job built on a computer and email. Ninety-six percent of respondents reported using email every day, 76 percent said they sit continually or almost continually, and 91 percent work indoors in a controlled environment every day.^[4] Twenty-seven percent reported daily phone conversations.^[4]

The work is also less solitary than the stereotype. Forty-three percent said they have face-to-face discussions with individuals and teams every day and another 48 percent at least weekly, and 30 percent rated working with a team as extremely important and 35 percent as very important.^[4] Contact with others is moderate, not constant: 56 percent said about half the time.^[4]

The activities O*NET rates most important are analyzing data or information, working with computers, processing information, and interpreting the meaning of information for others, in that order, with keeping knowledge current and making decisions close behind.^[4] In practice a day is split between hands-on work in code and data and explaining what it means, with accuracy mattering throughout: 40 percent said being exact is extremely important and another 40 percent very important.^[4]

Autonomy is real but bounded. Seventeen percent said they have a lot of freedom to set their own tasks, priorities, and goals and 79 percent some; 24 percent have a lot of freedom in decisions, 56 percent some, and 20 percent limited.^[4] Most respondents report some freedom rather than a lot, Competition is part of the setting: 65 percent described the job as highly competitive, and 52 percent meet strict deadlines weekly.^[4]

The largest employer is the corporate headquarters. In the federal wage survey, management of companies and enterprises employs 28,620 data scientists, more than any other industry, and computer systems design employs 27,590.^[1] BLS's broader grouping lists publishing, broadcasting, and content providers first at 12 percent of data scientists, followed by insurance carriers, computer systems design, and management of companies at about 10 percent each, and credit intermediation at 6 percent.^[3]

Where you work changes the job and the pay. BLS notes that asset management employers may want finance experience or coursework. Pay differs by industry too: software publishers pay a median of $156,220 while insurance carriers pay $107,680.^[1] The industry table in our salary guide lists each of the largest employers.

Geographically, the jobs concentrate: California (39,310), Texas (25,860), New York (23,970), Pennsylvania (13,810), and North Carolina (11,430) hold 43.6 percent of the 262,440 jobs in the wage survey.^[1] The deepest pools are the ones to search first, and the state table shows the job count and median for every state.

The best-paying state is not the biggest. Washington has the highest median among states with at least 1,000 data scientists, and after the Bureau of Economic Analysis's price adjustment it still leads, while California drops to 4th and lower-cost states such as Minnesota and North Carolina move into the top five.^[5] The cost-of-living section of our salary guide shows the adjusted figures.

1. Earn a bachelor's degree (about four years). BLS says data scientists typically need at least a bachelor's degree in mathematics, statistics, computer science, or a related field. Common fields of degree include mathematics, statistics, computer science, business, and engineering.^[3]

2. Build the mathematics and the code. BLS says the work involves algorithms and statistical techniques, so students need extensive study in mathematics and statistics, and suggests linear algebra, calculus, and probability and statistics in high school. At college level, computer science courses matter too, including data-oriented programming languages and statistical, database, and software tools for presenting analyses.^[3]

3. Apply what you learn to real data. BLS lists no related work experience as typical for entry, and does not name internships or portfolios; what the Handbook does say is that some employers want industry-related experience or coursework, such as finance coursework for a role at an asset management firm.^[3][2] Our reading is that projects and internships in the industry you want are the nearest substitute for experience.

4. Decide whether you need a graduate degree. BLS says some employers require or prefer a master's or doctoral degree.^[3] O*NET's survey is split between the two levels, covered under degree options below.^[4]

5. Apply to analyst and data scientist roles together. O*NET points to business intelligence analysts and clinical data managers as related occupations, so search those titles alongside data scientist.^[4]

Put together, the federal requirements describe a path that is a bachelor's degree, with no required work experience, or a bachelor's followed by a graduate degree. BLS gives no timeline for the occupation, so the length depends on the degree you choose.^[3]

The federal data lists no related work experience as typical for this occupation, so a bachelor's graduate with the right coursework can be hired directly.^[2] O*NET places the job in Job Zone Four, where a four-year degree is usual and a considerable amount of work-related skill, knowledge, or experience is needed.^[4] The two sources differ on experience, so expect postings to ask for more than BLS lists as typical, and treat projects as the way to close the gap.

The strongest entry evidence is work with real data. O*NET's most important tasks are analyzing, manipulating, or processing large sets of data with statistical software, creating graphs and charts, and testing and validating models.^[4] A student who has done each of those on a messy dataset, and can explain the results to a nontechnical reader, has done the central work of the job. The BLS point about industry-related coursework helps here: its example is finance coursework for a role at an asset management firm, and the same logic suggests pairing technical study with the subject of the industry you want.^[3]

Statistical modeling is also reachable from neighboring jobs. Computer systems analysts, operations research analysts, and software developers all list a bachelor's degree as the typical entry education, and the software developer median, $135,980, is above the data scientist's.^[2][1] Someone strong in code but weaker in statistics could start in a developer role with analytics projects, though that does not replace the mathematics the Handbook describes.^[3]

For business-minded students, a degree in business analytics is one route into the field. Our ranking analyzed 72 bachelor's programs. Check the curriculum for statistics, programming, and database courses, because those match the mathematics, programming, and database study BLS says the work needs.^[3]

O*NET scores each task for importance on a 100-point scale. The top of the list reads like a project: analyzing, manipulating, or processing large sets of data using statistical software (83); creating graphs, charts, or other visualizations (83); testing, validating, and reformulating models to ensure accurate prediction (81); delivering oral or written presentations of the results to management or other end users (80); and recommending data-driven solutions to key stakeholders (79).^[4]

Next come identifying business problems that data analysis can address (78), comparing models using statistical performance metrics such as loss functions (76), and cleaning and manipulating raw data (76).^[4] Further down are applying feature selection algorithms (69) and writing new functions or applications in programming languages (65), and reading scientific articles and conference papers to spot emerging analytic trends (60).^[4]

Two readings are worth drawing. First, tasks about communication and business problems score as high as the technical ones; presenting and recommending sit at 80 and 79, within a few points of the analysis itself. Second, writing new programs scores lower than using statistical software, so the job is closer to applied statistics with code than to software engineering. Someone who wants to build production systems should look at the pay gap with software developers and decide which job they actually want.

BLS names six qualities as important: analytical skills to research, examine, and interpret findings; computer skills to write code, analyze data, develop or improve algorithms, and use visualization tools; communication skills to convey results to technical and nontechnical audiences; logical-thinking skills to design and develop statistical models; math skills to use statistical methods; and problem-solving skills for the difficulties of data collection, cleaning, and modeling.^[3]

O*NET rates mathematics the most important skill (85) followed by critical thinking (81), and its work-style ratings add a personality profile. Attention to detail and intellectual curiosity both score 100, ahead of innovation (81), dependability (79), achievement orientation (71), integrity (70), and tolerance for ambiguity (66).^[4] The mix is telling: the job rewards people who are careful and curious, and who can stay with an unclear problem.

On tools, O*NET lists analytical software such as IBM SPSS Statistics, SAS, TensorFlow, and MATLAB; business intelligence and data analysis tools including Alteryx, Apache Spark, Google Looker, and Microsoft Power BI; cloud platforms including Amazon Web Services and Google Cloud; databases including MongoDB, PostgreSQL, Apache Hive, and Elasticsearch; and version control through Git, plus Docker, Kubernetes, and Apache Airflow.^[4] It also lists Microsoft Excel and PowerPoint, which is a reminder that spreadsheets and slides are still part of the job. No single tool is a requirement; the pattern is statistics software, a query language and database, a programming language, and a way to present.

The fastest way to build these is coursework with real projects, plus one or two end-to-end analyses of your own. A business analytics program or a statistics or computer science major provides the structure; the portfolio shows you can finish.

Bachelor's degree. This is the BLS entry credential, in mathematics, statistics, computer science, or a related field, including business and engineering.^[3] Our ranking of bachelor's in business analytics programs analyzed 72 programs on graduation rate, cost, and outcomes. If you are choosing a major, mathematics, statistics, and computer science map most directly to the BLS description; business analytics suits someone who wants the business side too.

Master's degree. BLS says some employers require or prefer one, and O*NET found that 44 percent of respondents said a new hire needs a master's, against 48 percent for a bachelor's.^[3][4] For someone whose bachelor's was in another field, a master's in a quantitative subject is one way to build the coursework employers look for. Our master's in business analytics ranking analyzed 89 programs.

MBA. An MBA is the management degree, with 978 programs in our ranking, and suits a data scientist heading toward a leadership role rather than a deeper technical one.

BLS does not publish pay by degree, so no federal figure shows what a master's adds. For an analytics path more broadly, our business analytics vs data science guide compares the two fields.

No license is required to work as a data scientist, and the Handbook names no required or recommended certification.^[3]

Software vendors and cloud platforms offer their own certificates, but BLS lists none for this occupation and does not publish pay by certification, so this page recommends none. If a certificate helps you learn a tool, such as a cloud platform that O*NET lists among common software, that is a fair reason to take one.^[4] Treat it as proof that you can use the tool, not as a substitute for the degree and the projects.

BLS says data scientists typically work in an office setting and that most work full time.^[3] O*NET's survey on hours is mixed: 63 percent reported a standard 40-hour week and 38 percent more than 40.^[4]

BLS does not report on remote work, so this page makes no claim about it. The pressure points are deadlines and accuracy. Fifty-two percent of O*NET respondents said they meet strict deadlines weekly or more, and the cost of a wrong number is high, with 80 percent saying accuracy is very or extremely important.^[4] Decisions carry moderate weight: 12 percent said their decisions have very important results for their employer, 28 percent important, and 44 percent moderate.^[4] The job suits people who like being exact and are comfortable being checked.

BLS does not publish a ladder for this occupation, and it does not give pay by years of experience, so the percentile range is the best federal guide to how far the job goes: $85,660 at the 25th percentile, $120,230 at the median, and $199,130 at the 90th.^[1] BLS does not describe what separates higher pay, but it does describe three specialties, engineering-heavy roles that build machine learning systems, research, and business strategy, and each is a different direction to specialize.^[3]

The management step is computer and information systems managers, who earn a median of $175,140; BLS lists 5 years or more of related experience as typical for entry, so it is a later move.^[1][2] Technical paths run through software development, with a median of $135,980, and statistics, with $105,650.^[1] For the business side of analytics, compare our business analyst career guide and business analyst salary guide.

BLS lists these related occupations: actuaries, computer and information research scientists, economists, financial analysts, market research analysts, mathematicians and statisticians, operations research analysts, software developers, and survey researchers.^[3] BLS groups them by similar duties, education, and pay.

Five occupations sit closest to this one on our comparison, and their economics differ. Over 2025 to 2035, BLS projects employment growth of 34.6 percent for data scientists, 11.9 percent for operations research analysts, 11 percent for statisticians, 10.2 percent for software developers, 7.9 percent for computer systems analysts, and 15.8 percent for computer and information systems managers, against 3.5 percent for all occupations.^[2]

Software development is the biggest and best-paid neighbor: 1,687,890 jobs at $135,980, with about 95,300 openings a year.^[1][2] It is the better choice for someone who wants to build and ship systems and is less interested in statistics. Statisticians earn $105,650 in a much smaller field of 29,030 jobs, and BLS lists a master's degree as typical entry education.^[1][2] Operations research analysts earn $88,940 and computer systems analysts $105,850.^[1] Each is a viable analytical career, but data science pays more than any of these three and is projected to grow faster than all of them.

See our salary guide for the full comparison table.

BLS treats AI as a source of demand for this occupation. It expects firms to continue integrating AI-based systems into their workflows and says data scientists will be needed to help businesses use AI and other technologies to improve processes, make informed decisions, design and develop new products, and market them.^[3]

That is not a guarantee for every task. The Handbook does not say which parts of the job tools may automate, and this page will not guess. O*NET's task list does show where the judgment lies: testing and validating models, identifying business problems that data can address, and recommending solutions to stakeholders all score high, while writing new programs scores lower.^[4] Our reading, not a BLS forecast: a data scientist who can decide which problem to model, check whether the output can be trusted, and persuade a manager is doing the part of the work the outlook expects to hold. One who only runs a template on clean data is doing the part most open to tools.

The median annual wage was $120,230 in May 2025, with the middle half earning between $85,660 and $158,880.^[1] Employment is projected to grow from 275,600 to 371,000 jobs between 2025 and 2035, about 24,800 openings a year including replacements. BLS attributes the growth to rising demand for data-driven decisions and to firms integrating AI-based systems.^[2][3]

The spread is wide. The lowest-paid 10 percent of data scientists earn less than $67,240 and the highest-paid 10 percent more than $199,130.^[1] Even the 10th percentile is $16,260 above the $50,980 median for all occupations.^[2] The salary guide breaks the pay down by state, industry, and cost of living.

The case for it. Pay is $69,250 above the all-occupation median, the field is projected to grow much faster than average, and a bachelor's degree with no related work experience is the typical way in.^[1][2] The work is varied, with specialties in engineering, research, and business strategy.^[3]

The case against. Entry is open in theory, but the work is competitive: 65 percent of O*NET respondents call it highly competitive, and 44 percent said a new hire needs a master's.^[4] BLS says raw data must be cleaned before analysis, and the title spans engineering, research, and strategy work, so a posting can describe very different jobs.^[3] Pay depends heavily on the employer, with a $48,540 gap between the best- and worst-paying large industries.^[1]

Who it does not suit. O*NET's interest profile scores Investigative at 100 and Conventional at 73, with Enterprising at 12 and Social at 11.^[4] Work styles rate empathy at 4, social orientation at 10, and leadership orientation at 12.^[4] Someone who wants to lead, sell, or work with people all day will find the day-to-day job quiet, and may prefer business analyst work or management.

The O*NET profile is a useful self-check. The strongest interest is Investigative, at 100: studying and researching, which for this job means questions answered with data. Conventional, at 73, covers organizing information, and the specific interest areas are mathematics and statistics (98) and information technology (80).^[4] The two work styles at 100, attention to detail and intellectual curiosity, are the personality match.^[4]

It suits you if you like finding out why numbers behave the way they do, can tolerate ambiguity (66 on the scale), and enjoy explaining results as much as producing them. It suits you less if you want to manage people early, or prefer creative or social work. If the technical depth worries you but you like data, business analyst roles are the nearer step, and our business analytics vs data science guide lays out the difference.