What is this career path?
Due to recent advances in technology, humanity is collecting vast amounts of data at an unprecedented rate. Data scientists use techniques drawn from computer science and statistics to generate actionable insights from this data. They work in industries as varied as medicine, online-education, finance and government, and have done everything from predicting the results of national elections to preventing house fires. In this profile we focus on the pros and cons of spending 2-5 years working as a data scientist.
Data science jobs involve obtaining data, cleaning it, analysing it and communicating actionable insights to decision makers in organisations. The title ‘Data scientist’ is used for a wide range of jobs, which range from entry-level positions only requiring basic knowledge of databases and Excel, all the way to high-level roles which involve inventing new algorithms, working with very large data sets and undirected research into open-ended questions. Entry level roles sometimes go by the name of ‘data analyst’ or ‘junior data scientist’. Read more on the different types of data science jobs here.
Data scientists are commonly employed by large tech companies, like Google and Facebook, tech startups, governments and non-profit research institutions (where they’re often just called analysts or researchers), e-commerce companies like Amazon and entertainment companies like Netflix.
Day to day, data scientists spend a lot of their time (often up to 80%) finding data sources and using computer programs to clean data to prepare it for analysis, running analyses and communicating their results by writing reports or doing presentations. They also spend time on generating project ideas and planning them out.
People who do well in data science roles are deep thinkers with intense intellectual curiosity. They are driven to learn new skills on their own and make new discoveries, often writing programs and analysing data in their spare time. They are also practically minded and happy to forego increasing the accuracy of their models once commercial goals are met. Working at a company whose product is personally fascinating is also important for success.2
For many entry-level roles, an undergraduate degree is sufficient. More advanced roles often require an advanced degree in a quantitative subject like computer science, physics, statistics or applied maths.
To get an entry-level job you usually need basic familiarity with a statistical programming language like R or Python, statistics, basic machine learning algorithms, database systems and data communication and visualisation. For more, see this skills checklist.
You can gain these skills either by self-learning and building a portfolio or by entering a data science bootcamp. Entry to Data Science bootcamps is highly competitive and placement rates vary. Our impression is that if you are smart in a quantitative and analytical way, dedicated and put a lot of effort into pre-study, you have a good chance of getting a bootcamp place.
Direct impact potential
As a data scientist your impact mainly comes from furthering the goals of the organisation you work for, meaning that your impact largely depends on where you work. Many organisations that employ data scientists contribute to innovation and are socially impactful, including the Center for Disease Control, Google, Khan Academy, Bayes Impact and DataKind. But even working for organisations whose goals aren’t directly linked with increasing social value may still be substantially valuable due to the “flow-through effects” of increasing general human empowerment. However your impact will be lower in industries with elements of zero-sum competition to gain market share such as marketing.
The US national salary ranges according to DataJobs:
|Data analyst (entry level)||$65,000-$110,000|
|Data analyst (experienced)||$65,000-$110,000|
According to O’Reilly, the major industries with the highest median salaries for data scientists are finance ($117k) and software ($116k). By country, the US has the highest median salaries ($117k), followed by Canada ($95k), Australia/New Zealand ($90k), and UK and Ireland ($82k).
Graduates of Zipfian Academy make an average $115K as a base salary following the program. Data Incubator reports that its graduates can get starting salaries of $100K to $200K.
You can predict your own expected salary by plugging variables into this regression model created by O’Reilly.
Data scientists are employed in a wide range of industries including technology, medicine and government, which gives you the opportunity to advise decision makers in the industry that you get a job in.
As a data scientist you develop skills in programming, machine learning, maths and statistics, as well as detailed working knowledge of the industry you work in. These skills are likely to become increasingly valuable in the future across a wide range of industries (hiring is projected to increase by 18.7%), which gives you great option value and strong bargaining power to negotiate salary and working conditions. The technical skills you develop also open the possibility to transition into different roles including software engineering, founding or joining early-stage startups, academia,1 and quant finance.
You can gain detailed working knowledge of the industry and organisation you work in, but otherwise data science isn’t especially good for exploration value.
Our impression from talking to data scientists is that the work is highly intellectually satisfying and the results of your work are highly tangible. Working conditions for data scientists are typically better than many corporate jobs, with 40-50 hour weeks (though this varies a lot on industry and team), and cultures of learning and mentorship are common. The biggest and most commonly cited downside of the job is cleaning data, which you spend a significant portion of your doing (but this varies by company and role), and is the least interesting part of the job.
|Focus is on actionable insights that further commercial goals and are implemented immediately, meaning there is faster feedback and a highly tangible connection between your work and outcomes||More emphasis on accuracy of models, exploratory and basic research, and knowledge for its own sake|
|No need to continually publish papers and apply for grants||‘Publish or perish’ culture, constant need to apply for research grants|
|No need to teach||Teaching is usually required|
|Can work in many parts of the world with relative ease||Far fewer jobs, so much less choice over location|
|Ability to communicate technical ideas to people with little to no technical background is essential||Less need for communicating technical ideas to non-experts|
|Strong domain expertise of the industry you work in is required, as well as good judgement of what will contribute to the goals of the organisation most efficiently||Strong expertise in your academic field is required, as well as a good sense of what research is currently ‘hot’ and will get cited|
- 4 types of data science jobs
Notes and references
- How easy it is to re-enter academia depends on the field. It is easier in computer science than in other fields.↩
- ”For many PhD holders, the key to success is to find a company whose product or service fascinates them, says Sebastian Gutierrez, author of Data Scientists at Work. “You need someone who is excited enough about the business that they actually care that they need to meet quarterly budgets and goals.”” Nature - Data science: Industry allure↩