Data analysis skills serve as an excellent starting point for anyone looking to become a data scientist. Find out how to transition from data analyst to data scientist in this guide.
With the current shift toward home working, many people are retraining in fields better suited to the 21st century economy. One field seeing major growth is data, with skilled data analysts and data scientists in huge demand.
Perhaps you’re considering a career in data and are keen to know what opportunities await you. Maybe you’re already working as a data analyst and want to know how you can progress into a data scientist role. The good news is that, although data analytics and data science denote two distinct career paths, data analysis skills serve as an excellent starting point for a career in data science. Once you’ve mastered data analytics, it’s a case of adding more complex and technical expertise to your repertoire—something you can do gradually as your career progresses.
So: How do you transition from data analyst to data scientist?
While there’s no single route into data science, this post outlines the main steps you’ll need to consider if you want to make the shift. Whether you’re a seasoned data analyst looking for a new challenge, or are new to the field and want to plan ahead, we offer a broad introduction to the topic.
- What’s the difference between a data analyst and a data scientist?
- Why become a data scientist?
- What additional skills do you need to learn in order to go from data analyst to data scientist?
- How to learn data science skills
- How to transition from data analyst to data scientist: Practical steps
- Key takeaways
Ready? Then let’s take a closer look.
1. What’s the difference between a data analyst and a data scientist?
Before you embark on your journey into data science, it can help to understand: What exactly is data science, and how does it differ from data analytics? First up…
What is data analytics?
Data analytics is the process by which practitioners collect, analyze, and draw specific insights from structured data (i.e. in a standardized format). Its ultimate aim is to inform decision-making. Although data analytics is a specialized role, it is just one discipline within the wider field of data science. You’ll find a more comprehensive explanation in this introductory guide to data analytics.
What is data science?
Data science is a much broader scientific discipline, of which data analytics is a single aspect. Data scientists generally work with large, unstructured (or unorganized) datasets. While a data analyst tends to focus on drawing conclusions from existing data, a data scientist tends to focus on how to collect that data, and even which data to collect in the first place. They need a far deeper level of insight into data than is required of a data analyst.
If this feels a bit vague, you can think of data science as being like the construction industry. Its purpose is to create data structures (like buildings) that can be used for specific purposes. Just as it takes many different skills to plan, design, and construct a brand new building, it takes many skills to plan, design, and construct these data structures.
Broadly, we can divide data science into the following categories, each with specific skill sets and tools associated with it:
Data theory, which involves creating entirely new, abstract algorithms. In our analogy, you can think of algorithms like bricks and mortar. While they’re useful for building things, somebody first had to invent them. Today, it is commonly accepted that machine learning is an invaluable data tool. But the person who invented the first machine learning algorithm had to show a great deal of foresight to understand its potential uses. Data theory is therefore highly technical. It is a skill (even, arguably, an artform) that not all data scientists have.
Data architecture involves taking algorithms and applying them to specific use-cases or fields, e.g. scientific or business domains. To continue our construction analogy, you can equate a data architect with a traditional architect. Their job is to combine algorithms (or bricks) in novel ways to create blueprints for specific types of data structures (or buildings)—just as a real architect would do.
Data modeling (or software engineering) involves taking an architect’s blueprints and figuring out how to put them into practice. Data modelers are like structural engineers. They take the raw material of algorithms and code and apply these to create software structures that are fit for purpose. They’re generally excellent coders since part of their job is to overcome unexpected hurdles and to fix things that don’t fit together as planned. These are the folk who get their hands dirty!
- Data analytics. Finally, you can imagine a data analyst as someone who uses the finished structure (or building) to do their job. For instance, if the structure was a fire station, you could think of a data analyst as a trained firefighter who knows how to use the different specialized aspects of the building. However, they won’t necessarily concern themselves with the intricate details of how the building was constructed.
As you can see, “data science” is really an umbrella term for a wide range of different disciplines. Another increasingly popular domain is data engineering—you can read about the difference between a data scientist and a data engineer here. So, if you’re thinking about a move from data analytics, consider which aspect of data science most interests you. This will help as you formulate a career plan. One thing’s for certain…whichever path you choose, you’ll have plenty to get your teeth into!
2. Why become a data scientist?
Considering the complexity of the field (and the fact that it takes a lot of time to gain the necessary skills) you might be wondering: Why become a data scientist? Here are a few reasons to consider moving into the field.
Data scientists are in demand
Demand for qualified and competent data scientists far outstrips supply. A 2018 study from LinkedIn showed that, in the US alone, there was a nationwide shortage of 151,717 data scientists. If you want a career where you’ll have no problem finding work, this is one to consider.
Data scientists get paid well
As you might expect for an in-demand role, data scientists tend to earn a pretty comfortable living. According to the salary comparison site Payscale, data scientists in the US earn around $67K to $134K per year.That’s a significant increase on data analysts, who usually earn between $43K and $85K.
As a data scientist, you add major value to a business
Since data analysts often focus on a single area (such as sales or marketing) they don’t always have full input into broader business strategy. That’s not true for data scientists, who are some of the most trusted members of the senior team. They’ll often sit on the Board, work directly with CEOs, and create strategic plans for the future of the business.
The field of data science is constantly evolving
Data scientists don’t have a single defined role. Since the position varies from business to business (and even from day to day) there are always exciting new problems to solve. Whether this means building brand new algorithms from scratch, creating data architectures, or just working in an area that’s completely novel to you, you’ll certainly never get bored.
Data scientists are needed in every industry
With data playing an increasingly important part in the economy, data scientists are needed in every industry you can think of. From healthcare to sports, finance, and e-commerce (not to mention the traditional sciences), the applications are almost limitless. For keen lifelong learners, this makes data science a cornucopia of opportunities to practice and grow.
3. Data analyst skills vs. data scientist skills
There are plenty of reasons to pursue a career in data science. But where to go from here? As a data analyst, especially a new one, you’re likely to be years away from a flourishing data science career. But this is good—it means you have plenty of time to develop your skills.
Start by conducting a “skills audit”—What data analysis skills do you currently have?
Before branching out, it’s advisable to carry out a personal audit of your data analytics skills. What gaps do you need to plug, and how can you go about filling them in?
Are you experienced using Python? What about R? Which programming language is better for pure analysis and which would you choose for application building? Do you have any experience working with relational databases like MySQL? What about collecting and cleaning data, manipulating it using MS Excel, or creating visualizations?
Don’t worry if you can’t answer all of these questions, but keep them in mind. In essence, you should aim to master your data analytics skills before progressing. We won’t get into detail here, but you can check out our guide to the key skills that every data analyst needs.
While practical skills can be learned, the most important soft skills to cultivate are:
- Critical thinking
- Analytical skills
- Presentation skills
So long as you nurture these core traits then you’ll have plenty to build on. It’s a long journey from fresh-faced data analyst to fully-fledged data scientist, and there’s no hurry. Every moment spent working as a data analyst counts as a valuable step in your journey towards becoming a data scientist.
What skills do you need as a data scientist?
In addition to being experts in data analytics, data scientists require an experimental mindset, a deep understanding of statistical methodologies, and a wide range of technical abilities. Which skills you require will depend a lot on your chosen career path or business domain. As a rough guide, you’ll need to develop at least some of the following abilities:
- Data languages, e.g. advanced Python and R (and others, if they relate to your field of interest).
- Relational databases, e.g. MySQL, PostgreSQL, Microsoft SQL Server, Oracle Database, SAP HANA.
- Machine learning algorithms e.g. Linear and logistic regression, decision tree, random forest, SVM, KNN, and more.
- Distributed computing, e.g. Hadoop, Spark, MapReduce.
- Data visualization, e.g. RShiny, Plotly, ggplot, Matplotlib (to name a few).
- Special skills, e.g. natural language processing (NLP), computer vision, optical character recognition (OCR), deep learning, and neural networks.
- API tools, e.g. IBM Watson, Microsoft Azure, OAuth.
- Postgraduate qualifications, such as a Master’s or Ph.D. in a field like computer science, statistics, or software engineering.
This is by no means an exhaustive list, but it does give you an idea of the skills you’ll need to develop. Whether you have a formal qualification or not, accumulating these abilities can take many years. That’s why you’ll need a natural passion for learning new things. If you see professional development as a tiresome necessity for career progression, this might not be the right career path for you. However, if you’re sold on the opportunities and want to move ahead, let’s explore how below.
4. How to learn data science skills
There’s no sugar-coating it: The process from data analytics to data science is gradual and often imprecise. This can be challenging but also be rewarding, as it means you can carve your own career path. The first step is to take charge of your personal development. Pursuing your interests will help you build the foundational skills you need, while allowing you to decide which areas of data science most interest you. While the transition won’t happen overnight, the good news is that you can start right away.
Learn some new programming languages (and other technical skills)
Most data analysts get by with a solid understanding of Python. Data scientists usually add the programming language R to their arsenal, too. Check out someintroductory tutorials for R, or advance your Python skills by building applications in your spare time. Whatever you do, challenge yourself—you’ll learn best by experimenting and making mistakes. Aim to upskill in other technical areas as well, for instance by playing around with distributed computing or statistical tools.
Dabble with machine learning (and other) algorithms
Using existing tools is one thing. However, data scientists often have to create solutions from scratch. Machine learning algorithms are a common example, and are often used in data science. Dabble with algorithms like decision trees or random forest to get a feel for how they work. Read around the topic and you’ll learn which ML algorithms work best for different data types, and which tasks they can be used to solve.
Follow the latest news and events in the field
For a broader feel of what data science offers, follow industry thought leaders on social media, or subscribe to some publications. This won’t just help you get a better overall picture of the field (including things like data architecture and modeling) but will also expose you to the latest developments. If you’re on Twitter, check out Andrew Ng, Kirk Borne, Lillian Pierson, or Hilary Mason, for starters.
Get on GitHub
A data scientist who’s not sharing projects on GitHub is like a baker without bread! Many companies and organizations use GitHub for version control and for sharing code. It’s important, then, that you actively use it. Why not share some projects? If you’re in need of some inspiration, you’ll find a collection of unique data project ideas in this guide.
Enter some Kaggle competitions
Kaggle is a great place to practice your data science skills in a safe, web-based environment. They offer regular, practical tasks where you can get to grips with data modeling, machine learning, and more. Don’t fret about doing a perfect job. As we said above, you learn by making mistakes. Aim to fail forward. Once you’re feeling confident, why not find a dataset online and have a go on your own?
5. How to transition from data analyst to data scientist: Practical steps
Learning the necessary skills is a great place to start. However, the bigger challenge is having the confidence to make your ambitions known. After a few years in data analytics (building your knowledge as we’ve described above), you may find that you’re ready to pursue a more formal route into data science. Here are some practical tips for how to proceed:
Take a structured course
While it’s great to explore different tools and skills, it’s a good idea to cement what you’ve learned through a structured data science course. While there’s no substitute for working on real projects, there’s no harm in getting an online qualification, either. It’ll look good on your resumé and will show any potential employers that you’re serious about moving into the field.
Make a list of companies you’d like to work for
Which companies inspire you? Think about those you’d love to work for and write them down. Don’t limit yourself—aim high. Add to the list as new companies catch your eye. Once in a while, check out their data scientist job listings (specifically, the skills section) and make a note of what you’re missing. This is great for deciding which new skills to focus on.
Create a data science portfolio
Whether you’re already working as a data analyst or aspiring to be one, you should have—or be in the process of building—a professional data analytics portfolio. As you gradually expand your skillset to include data science, you can reflect the transition in your portfolio. For example, once you’ve done a few Kaggle projects and put them on your GitHub, update your portfolio. Create a couple of case studies, share some articles you’ve found interesting or even ones that you’ve written yourself. By channeling your pet projects and personal interests into one place, you’ll have something tangible to share with employers. Even if you haven’t formally worked in data science before, this will show them that you’re serious about it.
Do what you can to get noticed at work
Why not volunteer to run a lunch and learn training session at your office? Or even organize a company hackathon? The business you work for might not currently employ many (or even any) data scientists but there’s nothing like showing a bit of initiative to demonstrate your value. Make a good impression at work and you never know when it might come back around—even if it’s just in the form of a glowing recommendation to a future employer.
Apply for jobs, even if you don’t think you’re ready for them
Seen a job that looks appealing, but only have some of the skills required? Apply anyway. Many skills are listed as “desirable” not “essential”, which means you may still stand a chance. Even if you get rejected, you’ll learn something new every time and you’ll come away with a better sense of what organizations are looking for. Plus, if you keep applying for jobs at your dream company, they might start to remember you. Persistence pays off.
Build your network
As the old saying goes: it’s not what you know, it’s who you know. While “what you know” is certainly important in this case, so is building a network. Talk to other data scientists, connect with people whose projects you admire, and attend industry events. You’ll be surprised how much people are willing to help if you need it. And when it comes to applying for that first job, who knows? Maybe you’ll find it through your network.
6. Key takeaways
As we’ve seen, data science is not so much a single career destination as a journey in personal development. While the fact that there’s no single path into data science can be a challenge, this is also what makes it such a diverse, fascinating, and rewarding field to work in. If you’re curious, open to experimentation, analytically-minded, and love learning new things, then a career in data science might well be for you.
Indeed, data science is not for everyone. There’s no overnight path to success, and it requires the accumulation of plenty of technical expertise. However, it’s an ideal next step for those who have started in data analytics and want to invest in their future career. Dip a toe into data science today, and who knows what the future holds?
Are you yet to get started with data analytics? Try this free, five-day data analytics short course. Meanwhile, to learn more about where a career in data analytics can potentially lead you, check out the following posts: