from data analyst to data science is a viable way to break into the data science field, and this article aims to explain how you can make that transition.
Why be a data analyst first?
I often recommend becoming a data analyst first and then transitioning to a data scientist.
Now, why do I do this, given I have never worked as a data analyst? Well, it’s for the following reasons.
- Becoming a data analyst is easier than becoming a data scientist.
- You truly learn and understand the business impact data can have — beginner data scientists often focus on building fancy models instead of solving business problems.
- At some companies, you may even do the same job as the data scientist despite the title differences.
- Time in beats timing. So, being in the industry is always better in my opinion.
A comprehensive roadmap to becoming a data analyst is beyond the scope of this article, but I’d be happy to create one if that’s something that interests you.
What is the difference between data analyst and scientist?
Even though data analysts and scientists can be similar at some companies, the roles do differ in most cases.
In general, a data analyst is more business decision-focussed and will work with tools like:
A data scientist will pretty much be able to do everything a data analyst can and will have more advanced abilities in:
You can think of it as data analysts are more concerned with looking at what happened, and data scientists are more concerned about what will happen, e.g. predicting the future.
You don’t have to transition to data science from data analytics; I know many people who are fantastic analysts and are happy in their current role, getting a lot of fulfilment and being compensated very well.
However, I also know many people who want to move to data science and are using the data analyst position as a stepping stone.
Neither is right or wrong; it just comes down to what your goal is. Chances are, if you are reading this article, then you want to make the jump, so let’s go over why becoming a data analyst first is not a bad thing at all.
Skills to develop to transition
To move from data analyst to data scientist, you need to learn the following.
Maths
If you are working as a data analyst, you likely already possess decent statistics skills, so the primary areas you need to focus on are linear algebra and calculus.
- Differentiation and the derivatives of standard functions.
- Partial derivatives and multivariable calculus.
- Chain and product rule.
- Matrices and their operations, including features such as trace, determinant, and transpose.
Coding
As a data analyst, your SQL skills are probably already excellent, so the main thing you need to improve is Python and general software engineering.
- Advanced Python concepts like unit testing, classes and object-oriented programming.
- Data structures and algorithms, and system design.
- An understanding of cloud systems like AWS, Azure or GCP.
- ML libraries such as scikit-learn, XGBoost, TensorFlow, and PyTorch.
Machine learning
You don’t need to be an ML expert, but you should understand the basics pretty well.
How to learn?
Self-study
The most straightforward and intuitive approach is to study in your spare time, either after work or on weekends.
Some people may not like that, but if you want to make a change in your career, you need to put in time and effort; that’s the brutal truth. Loads of people want to be data scientists, so it’s no walk in the park.
There are numerous resources available to learn about the above topics, and I have written several blog posts on the exact books and courses you should use.
I will leave them linked below, and I highly recommend you check them out!
The pros of self-study are:
- Very cost-effective and can even be completely free.
- Learn on your own schedule.
- Custom learning path.
And the cons:
- There are no clear structures, so it’s easy to go wrong.
- No formal credentials.
- Requires high discipline and motivation.
Degrees
You can always return to school and pursue a formal degree in data science or machine learning.
The pros of this approach are:
- Emphasis on mathematics, statistics, computer science, and algorithmic understanding.
- A degree (especially from a top university) carries more weight with some employers.
- Access to faculty, alum networks, research projects, and internships.
The cons are:
- It may be too theory-heavy and lacks real-world projects and data.
- Takes 2–4 years (Bachelor’s) or 1–2 years (Master’s).
- Can be expensive
- Need strong academic record, possibly GRE, letters of recommendation, or prerequisite coursework.
Bootcamps
These have emerged everywhere in recent years due to the growing demand for data and machine learning roles.
In general, they offer a cheaper alternative to degrees, with more hands-on projects and practical lessons.
The pros are:
- Most boot camps are 3–6 months long, focusing only on data science skills.
- Heavy focus on real-world projects, coding, and tools (Python, SQL, machine learning libraries).
- Many offer career coaching, resume reviews, mock interviews, and job placement support.
- Cheaper than a degree.
And the cons:
- Shallow theoretical depth.
- It can be too fast-paced.
- Quality can vary, so be sure to do your research before participating.
- Limited credibility to employers.
At your current job
This is my favourite, and it’s the most effective and worthwhile.
You can learn everything in your current job if you work on the right projects and also express interest to your manager about the skills and tools you want to develop.
Managers love it when their direct reports take the initiative and show passion for their work because it also benefits them as a byproduct.
The pros are:
- Getting paid to learn, what a win!
- Access to real-world data and business problems.
- Real life data science experience to add to your portfolio.
- It might even allow you to transition full-time to data science.
The cons are:
- This could lead to more workload.
- Role expectations may be fixed, and there may be little to no internal mobility.
Creating your portfolio
During and after your studies, you need to create some evidence of the work you can do as a data scientist, basically making a portfolio.
I’m planning to release a more in-depth video soon on what a strong data science portfolio should include. But for now, here’s the short version:
- Kaggle competitions — Do one or two. It’s not about placing high; it’s about showing you can work with real datasets and follow through.
- 4–5 simple projects — These should be quick builds you can complete in a day or two. Upload them to GitHub. Even better, write short blog posts to explain your process and decisions.
- Blog posts — Aim for around five. They can cover anything data science-related: tutorials, insights, lessons learned — just show that you’re thinking critically and communicating well.
- One solid personal project — This is your centerpiece. Something more in-depth that you work on over a month, an hour or two each day. It should showcase end-to-end thinking and be something you’re genuinely interested in.
That’s it.
People overcomplicate this step way too much. Just start building — and keep showing up.
Getting the job
As I said above, the easiest way is to transition internally.
If this is not an option, then you need to get busy applying!
You need to align your CV/resume, LinkedIn profile, and GitHub account with the data scientist job role. Ensure you start referring to yourself as a data scientist, not “aspiring.”
I studied physics at university, but I have never been paid to practise physics; I am still a physicist. The same applies to data science.
Utilise your portfolio everywhere you can to demonstrate your abilities. Your GitHub profile should link to your LinkedIn profile, which should then link to your blog posts and other relevant content. Get an ecosystem that traps people so they “spend” more time with you.
After everything is sufficiently prepared, start applying for more analytics-focused roles with the title data scientist. You can, of course, go for the more machine-learning ones, but they will be harder to get.
Leverage your network as well for referrals. If you have been working in the data field for some time, there must be at least one person you know who can refer you to a data science job.
The beauty of transitioning from a data analyst to a data scientist is that you can take your time, as you are already earning money and in the field, which takes the pressure off. Just make sure you stick to it and make consistent progress!
Another thing!
I offer 1:1 coaching calls where we can chat about whatever you need — whether it’s projects, career advice, or just figuring out your next step. I’m here to help you move forward!
1:1 Mentoring Call with Egor Howell
Career guidance, job advice, project help, resume reviewtopmate.io
