You’re passionate about big data. You’re ready to commit to an advanced degree. But should it be the MS or PhD? Our guide to doctoral programs in data science is here to help. It has advice on benefits & downsides, job opportunities, dissertation topics, courses, costs, and more. Just want the schools? Skip ahead to our complete list of data-related PhD programs.
Why Earn a PhD in Data Science?
A PhD in Data Science is a research degree designed to give you a deep-rooted knowledge of statistics, programming, data analysis, and subjects relevant to your area of interest (e.g. machine learning, artificial intelligence, etc.).
The key word here is research:
- You’ll be expected to conduct your own experiments in a specific field.
- You’ll focus on theory—both pure and applied—to discover why certain methodologies are used.
- You’ll take apart tools & technologies to determine how they’re built.
Benefits vs. Downsides
The benefits and downsides of a PhD are always being debated, and Quora has plenty of insightful answers from data scientists on this subject! Here are some of our findings…
Benefits of a PhD in Data Science
In a PhD program, you’ll have the opportunity to:
- Research an area in data science that may a) be about to the change the industry b) have unexpected applications c) solve a long-standing problem.
- Collaborate with “star” academic advisors in well-funded data science institutes and centers.
- Become a critical thinker—knowing when, where, and why to apply theoretical concepts.
- Specialize in an upcoming field (e.g. biomedical informatics).
- Gain access to massive, real-world data sets through university partnerships.
- Work with cutting-edge technologies and systems that may not be available in the private sector.
- Automatically earn a master’s degree on your way to completing a PhD.
- Qualify for high-level executive or leadership positions.
Downsides of a PhD in Data Science
On the other hand, you should realize that a PhD program:
- Takes 4-5 years on a full-time schedule to complete. These are years when you could be earning money and learning real-world skills in a major company.
- Can be prohibitively expensive if you don’t find ways to fund it.
- May focus too much on pure theory and not enough on pragmatic applications & practical experience.
- Contains a lot of solitary hours in reading and writing. If you want to work with a team or collaborate with industry partners, choose your program & dissertation topic very carefully.
- Won’t give you “on-the-job” knowledge of problems and demands in the corporate world.
As a former PhD student (albeit not in data science), the most important piece of advice I can give you is to pick the right advisor. This is the person who will guide your research, help you with funding, connect you to opportunities & resources, keep you on track, and launch your career. Make people a top priority when considering programs.
Do You Need a PhD to Land a Job?
In most cases, the answer is “no.” In the past, companies had a habit of demanding a PhD in Statistics or Computer Science for top-level positions. This was often because there weren’t many master’s programs in data science and employers wanted reassurance that candidates knew their stuff.
That situation has changed. Because master’s programs now emphasize hard skills & real-world applications, the number of job listings requiring a PhD is dropping rapidly.
Our best advice is to pay attention to the employer and the job title:
- Companies & labs that specialize in data science, and major tech players like Amazon and Facebook, will have a reason for specifying a PhD in the education requirements.
- Other industries may be happy with a strong BS or MS degree and relevant work experience. If you know how to analyze results, make product decisions, increase efficiencies, predict trends, and provide insights, you’re in good shape.
Have a look at the debate on Quora if you’d like more perspective on this question.
Careers for Data Science PhD Holders
PhD holders find careers in academia, industry & university research labs, government departments, and big-name tech companies. These are places that need job candidates who can:
- Research & develop new methodologies.
- Build core products, tools & technologies that are based on data science (e.g. ML or AI algorithms for Google or the next generation of big data management systems).
- Reinvent existing methods & tools for specific purposes.
- Translate research findings & adopt theory to practice (e.g. evaluating the latest discoveries & finding ways to implement them in the corporate world).
- Design research projects for teams of statisticians and data scientists.
Sample job titles include:
- Director of Research
- Senior Data Scientist/Analyst
- Data/Analytics Manager
- Data Science Consultant
- Laboratory Researcher
- Strategic Innovation Manager
- Tenured Professor of Data Science
- Chief Data Officer (CDO)
Typical Program Structure
Data science PhDs are built on the same structure as most doctoral programs. That means you’ll typically have to:
- Complete 2 years of full-time coursework.
- Pass a comprehensive exam—often both oral and written—that shows you have mastered the subject matter.
- Submit a dissertation proposal and have it approved.
- Devote 2-3 years to conducting independent research & writing a dissertation. You’ll probably be teaching undergraduate classes at the same time.
- Defend your work in a “dissertation defense”—usually an oral presentation to academics and the public.
During these years, you can improve your career prospects by attending and speaking at conferences, applying for summer fellowships, consulting, and/or doing paid work for employers (e.g. part-time research).
PhD students are expected to make a creative contribution to the field of data science—that means you’re not allowed to go over old ground or rehash what’s already out there. Your contribution will be summed up in your dissertation, which is a written record of your original research.
Some students go into a PhD already knowing what they want to research. Others use the first couple of years to explore the field and settle on a dissertation topic. Your advisor will be your closest ally in this process – one of the many reasons to choose him/her wisely!
Data Science vs. Business Analytics vs. Specialties
Doctoral programs in data science can also fall under the heading of Statistics, Computational Sciences, Informatics, and the like. Because there’s really no rhyme or reason to the title, evaluate each program on its curriculum instead. Make sure the foundation courses and electives will prepare you for the research area that you want to explore.
The exception to this rule? The PhD in Business Analytics (or Decision/Management Sciences). These programs are typically administered through the School of Business, which means the curriculum often includes corporate topics like management science, marketing, customer analytics, supply chains, etc.
Interested in a particular subset of data science? Some universities are developing specialty PhD programs. Biostatistics and Biomedical/Health Informatics are obvious examples, but you’ll also find a number of doctoral programs in Machine Learning (usually run by the Department of Computer Science) and sub-specialties in fields like Artificial Intelligence and Data Mining.
As always, look for super-smart advisors and find out what’s happening in their research labs!
Typical Admissions Requirements
PhD candidates have to fill out a lengthy application form and pay an application fee. Universities usually want to see applicants who have:
- A Bachelor of Science (BS) in Computer Science, Statistics, or a relevant discipline (e.g. Engineering) with an official transcript from an accredited institution
- A GPA of 3.0 or higher on a 4.0 scale
- GRE test scores
- TOEFL or IELTS for applicants whose native language is not English
- Letters of recommendation
- Statement of purpose/intent
- Résumé or CV
If you don’t already have certain skills (e.g. stats, calculus, computer programming, etc.), the university may ask you to complete prerequisite courses.
In the 2017-2018 academic year, many universities were charging between $1,300-$2,000 per credit hour. With a PhD of 70-75 credits, you could be looking at tuition costs that exceed $150,000. On top of that, you’ll have to pay various fees (e.g. lab, health insurance, technology, international student, etc.).
That’s the bad news. The good news is data science is a high-demand field, so almost all universities will help you cover tuition & living costs. The PhD in Big Data program at Brown, for instance, guarantees five years of financial support—including a stipend and health insurance—to students in good academic standing.
One last thing to consider is your “cost of living” expenses (e.g. housing, books & supplies, food, transport, conference travel, etc.). Does the university provide subsidized housing for PhD students? Can you cover rent with your stipend? Do you mind eating packaged noodles? Use Sperling’s Best Places to compare costs in multiple cities.
How to Pay for a PhD
Your first stop should be the PhD program page on the university website. Here you’ll find links to relevant fellowships and advice on financial matters, including:
- PhD Fellowships: These are scholarships by any other name and they’re usually service-free (i.e. you don’t have to work for them). You’ll find lots of fellowships sponsored by the university, by companies or industries, and by the government (e.g. National Science Foundation). Be aware that some external fellowships will only cover the years of your dissertation research.
- Teaching/Research Assistantships: Assistantships are a popular way for universities to fund PhD students. In return for teaching undergraduates or working as a researcher, you’ll often receive a break on tuition costs and a living stipend.
- In-State Tuition: Public universities (e.g. University of North Carolina) usually offer in-state students a much lower cost per credit. This can be a real boon if you live near a strong school.
- Regional Discounts: Many universities have agreements to offer reduced tuition costs to students from neighboring states (e.g. New England Board of Higher Education Regional Student Program (RSP)). Check to see if this applies to your PhD.
- Travel Grants: It’s important for doctoral students to attend research conferences and network with future collaborators, and a lot of grants are designed with this purpose in mind.
- Student Loans: Consider all your other options before going into debt! A doctorate is a long-term commitment—you may not see a financial return on your education investment for good number of years.
Remember, a lot of PhD students in data science are fully funded. Just to take a few examples:
- Most students in Johns Hopkins University’s PhD in Machine Learning are fully funded through a mixture of research assistantships, teaching assistantships, training grants, and fellowships.
- U.S. citizens and permanent residents in Stanford’s PhD in Biomedical Informatics are funded by a National Library of Medicine (NLM) Training Grant and Big Data to Knowledge (BD2K) Training Grants.
- Warwick University in the U.K. offers approximately 10 fully funded 4-year scholarships to U.K. residents and eligible E.U. students.
If you’re coming from overseas, be sure to talk to your school about any differences between funding for citizens and international students.
Online vs. On-Campus Data Science Programs
Distance education is tempting if you can’t afford to move, and a small number of universities now offer online PhDs in data science. These programs often require you to attend a few campus events (e.g. symposiums), but allow you to complete coursework and conduct research in your own hometown.
We advise you to think very carefully about online learning. It may be a great option if you’re a seasoned pro in San Jose with access to state-of-the-art facilities, fabulous research sources, and on-the-ground mentors. It may not be so great if you live in rural Wyoming.
6 Questions to Ask Yourself Before You Commit
- Are you extremely passionate about an area of research?
- Do you know which academic(s) you want to work with? Are they willing to work with you?
- Do you mind committing to 4-5 years of study when you could be working full-time instead?
- Does your university have strong funding sources (private & government) for data science research?
- Will you have access to super-cool data resources, labs, and industry partners?
- Do you know how you’re going to pay for it?
If you’ve answered “no” to any of these questions, you may wish to reconsider your decision. Trust us, it’s going to be a long haul!
We found 47 universities in our directory offering doctorate-level programs in data science. If you represent a university and would like to contact us about editing any of our listings, or adding new programs, please send an email to info (at) mastersindatascience.org.
PhD in Data Science/Analytics Online
Click here to see the full list of on-campus programs instead.
Colorado Technical University
Students in Colorado Technical University's program leading to a Doctor of Computer Science with a concentration in Big Data Analytics can complete most of the coursework online. However, during their course of studies they must attend at least four residential symposiums at the campus in Denver, which are offered four times per year. First-term students must also attend an on-campus orientation, which is scheduled in conjunction with a symposium. The three-year program requires students to complete 96 credits, and to research, write and defend a dissertation. Applicants should have a master's degree with a GPA of at least 3.0, although CTU also offers the option of earning an accelerated master's degree while starting work on the doctoral degree in a program called the Doctoral Advantage.
Indiana University Bloomington
Doctoral students at Indiana University Bloomington can minor in Data Science, gaining skills that are valuable in fields such as education, business, environmental science, public heath, political science, and sociology. In data science classes, student learn to analyze, visualize and report on large amounts of data. To achieve the doctoral minor in data science, students must complete at least 12 credits of data science coursework, selected in consultation with an adviser from the data science program. Candidates must complete each data science course with a grade of B or better. There is no requirement for a written qualifying exam in a minor field.
Johns Hopkins University
University of North Texas
PhD in Data Science/Analytics On-Campus
Click here to see the full list of online programs instead.
Arizona State University
California Institute of Technology
Carnegie Mellon University
Columbia University in the City of New York
George Mason University
Georgia State University
Indiana University-Purdue University-Indianapolis
Iowa State University
Johns Hopkins University
Kennesaw State University
New York University
Oregon Health & Science University
The University of Tennessee
The University of Texas at Austin
Tulane University of Louisiana
University at Buffalo
University of Arkansas at Little Rock
University of Cincinnati
University of Illinois at Chicago
University of Iowa
University of Kansas
University of Maryland-College Park
University of Massachusetts-Boston
The College of Management at the University of Massachusetts Boston offers a Doctor of Philosophy in Business Administration degree with an information systems for data science track. While it is possible for students to request some required courses to be waived based on prior studies, all students must complete at least 14 courses to be eligible for graduation. The first two years of the program consists of coursework; the remaining time of enrollment is spent on research related to a student's dissertation topic. PhD assistantships are available for qualified doctoral candidates. Assistantships include full tuition and fee waivers in addition to a yearly stipend of $25,000 for three years. Most courses meet once a week during late afternoon and evening hours.