How to Switch from Data Analyst to Data Scientist

Are you a Data Analyst looking to break into data science? If so, this post is for you.

Many people start in analytics because it generally has a lower barrier to entry, but as they gain experience, they realize they want to take on more technical challenges, dive deeper into machine learning, or even just increase their earning potential. Moving from Data analyst to Data Scientist can be a smart career move — but it requires the right strategy.

If you’re new here, my name is Marina. I’m an Applied Scientist at Amazon, and I’ve helped dozens of people transition into tech, even from non-traditional backgrounds — myself included.

In this post, we’re going to cover everything you need to know to make the transition from data analyst to data scientist successful:

What skills you’ll need to develop

My favorite learning resources

And strategies for landing interviews and securing job offers

Let’s get into it, starting with deciding if this transition is even a good idea for you in the first place.

Role comparison

Before we get started, let’s just make sure we’re all on the same page about what the difference is between these roles anyway, starting with data analytics.

Data analysts focus on working with structured data to drive business decisions. Their toolkit typically includes SQL, Excel, Tableau or PowerBI, and basic Python for data processing, visualization, and maybe simple statistical analyses. The role centers on understanding historical data to answer questions about what happened and why.

Data scientists build on these foundations but extend into predictive modeling and automated decision-making. While they also use SQL and Python, they work more extensively with statistical modeling, machine learning frameworks, and cloud platforms. Their focus shifts to predicting future outcomes and recommending actions.

A common misconception is that data analysts must become data scientists to advance their careers. That’s definitely not true!

Senior analysts can earn high salaries and have a really strong business impact without deep ML or statistical knowledge.

Honestly, not everyone is going to enjoy data science work, and many would be happier staying on the analytics path.

So before we go any further, ask yourself the following questions:

Are you curious about machine learning and how it works?

Are you comfortable with (or at least interested in) advanced mathematics and statistics?

Are you comfortable with technical challenges and software engineering concepts?

Are you ok with a role that has a lot of ambiguity, both in the daily work and the career progression?

If you’re still with me and thinking “Yes, I definitely want to pursue data science,” let’s talk about how to make it actually happen.

Skills needed to transition

Alright, so now that you’ve decided to make the transition, let’s break down the key skills you’ll need to develop. We’ll focus on four core areas that form the foundation of data science work.

Mathematics & statistics

If you’re coming from an analytics background, you probably have some exposure to statistics, but data science might require a bit more depth on the math front. You’ll need to be comfortable with:

Multivariable calculus and linear algebra, particularly matrix operations and gradients for understanding machine learning algorithms. But don’t worry — you don’t need to be a math expert, you just need enough to understand the fundamentals to help you grasp how algorithms work.

You’ll also need probability theory and hypothesis testing for experimental design.

As well as statistical concepts like different types of distributions and regression techniques

And ideally, some experience with causal inference

Programming

If you’re already using SQL and basic Python in your role, you have a head start here. Now it’s just about leveling up. Focus on:

More advanced Python, so things like OOP fundamentals, writing modular maintainable code, unit tests, performance optimization, and so on.

Using ML frameworks like scikit-learn, Tensorflow, and PyTorch.

And familiarity with basic data structures and algorithms for coding interviews. Generally this will just be questions on arrays and strings, so you don’t need to go too crazy with this, but it’ll be important to know for interviewing.

Machine learning & AI fundamentals

This is another core pillar of data science, so you’ll want to be comfortable with ML fundamentals like:

Supervised learning (so, regression and classification).

Unsupervised learning (things like clustering and dimensionality reduction).

Model evaluation and validation.

Deep learning basics.

And these days, being familiar with GenAI is a plus (but by this I mean learning how to work with APIs, not training models from scratch)

Big data & data engineering concepts

Finally, many data science roles involve working with large-scale datasets and building automated pipelines. For this, you’ll want to focus on:

Working with cloud computing platforms, particularly AWS services like S3 and SageMaker

Data pipeline development using tools like Airflow

Potentially basic system design principles for scaling your solutions (this is more important as you become more senior or focus more on ML).

How to develop these skills

Now that we’ve covered what you need to learn, let’s talk about how to actually build these skills. There are a few different paths you can take, and the right one for you will depend on your budget, learning style, and schedule.

Self-study

If you’re self-motivated and disciplined, self-study can be a totally reasonable and cost-effective way to transition into data science. The key is consistent practice and choosing the right resources.

Here are some great courses I’d recommend checking out, in order (these are affiliate links, btw!):

Start with the DeepLearning.AI Mathematics for Machine Learning and Data Science Specialization (affiliate link!). This covers linear algebra, calculus, and probability and statistics, so you’ll have a solid foundation for the next things to learn. It’s presented in a really accessible way, so don’t be scared about starting with math.

Then, check out the classic Stanford Machine Learning Specialization (affiliate link!), which gives a solid introduction to ML fundamentals.

After that, the DeepLearning.AI Deep Learning Specialization (affiliate link!), at least the first three courses.

And finally, as a bonus I’d recommend the Generative AI with Large Language Models course (affiliate link!). This is a short course that is a good way to get familiar with modern AI applications.

You’ll also need to get an understanding of basic DSA for coding interview prep. For this I enjoyed Educative’s Grokking the Coding Interview Patterns in Python, which focuses on common patterns for data structures and algorithms questions. I found this really helpful so that it doesn’t just seem like you need to “know the trick” to answer the LeetCode problem.

And, a few books that are worth reading (these are also affiliate links, but I do <3 all these books):

Practical Statistics for Data Scientists — A great resource to deepen your statistical knowledge along with examples in R and Python.

Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow for practical ML examples.

Software Engineering for Data Scientists — This book is great and can help bridge the gap between the kind of coding you might be familiar with for analytics and production-level coding.

There are tons more, but this would be my top three. Here is a link to more of my favorite technical books if you want to explore further!

The most important thing when going the self-study route is consistency. Make a schedule and stick to it, even if it’s just a little bit each day.

Bootcamps

Now, maybe you’re thinking you’d prefer to have a little more structure and outside accountability in your learning. If you don’t want to commit to a full degree, bootcamps can be another option.

Some pros of bootcamps are:

Fast-paced learning — You generally can complete them in a few months.

Structured curriculum, because everything is laid out for you, so you don’t have to piece together your own learning plan.

And community support — You get to learn alongside peers and get mentorship from instructors who may be folks already working in the field.

One thing to keep in mind is that bootcamps vary in quality, and not all are super valued by employers. Before enrolling, do your research — so, check reviews, talk to alumni, and make sure they offer career support.

Master’s degree

For those looking for a deep dive into data science with strong networking opportunities, a Master’s degree can be a solid investment. This is especially useful if you’re transitioning from a non-technical background, or if you’re worried your background won’t be passing resume scanning tools.

The downside is obviously that Master’s programs can be expensive and time-consuming. But the good news here is that there are now affordable, part-time online programs that allow you to study while working. For example, Georgia Tech’s programs are really affordable and of pretty decent quality.

Mentorship

No matter which path you take, mentorship can be incredibly helpful. Having someone to guide you, provide feedback, and help with career navigation can make a huge difference.

Some ways to find mentors:

At your company — If your company has data scientists, ask if you can collaborate or shadow them.

LinkedIn — Join data science groups or reach out to professionals (I have a whole video on mentorship strategies if you need help with this!).

Online communities like Reddit, Discord servers, and Slack groups can be another avenue to connect with fellow learners and professionals.

Or, hire a mentor — If you’re serious about leveling up quickly, investing in a mentor can be worth it.

Demonstrating experience

Ok, so you’ve learned all the skills you need. That’s great, but how do you prove to a potential employer that you actually can do the job of a Data Scientist?

I have a whole video on how to build a portfolio and get experience outside of your full-time employment. The TL;DR there is that you should try your best to do self-motivated projects that allow you to simulate the working conditions of being on the job as closely as possible.

But if you’re reading this post, there’s a decent chance you’re currently working as a Data Analyst already, which gives you a whole other set of opportunities to leverage within your current role.

For example, let’s say you’re regularly creating reports in Excel or Tableau. You could automate this process with Python scripts, maybe even add some predictive elements. Or if your company runs A/B tests, volunteer to help with the statistical analysis.

If you have a data science team, try to collaborate with them on a project. And if there isn’t a data science team, pitch your employer on some impactful projects that would also help you to learn.

Best case scenario, this can result in an internal transition. Worst case, you now have concrete examples of impact and real data science projects to include on your resume.

Getting a job

If you’re able to transition internally then great, you’re done! If not, here are some strategies to help you get that first Data Science role:

First, let’s talk about how to position yourself online. Your resume, LinkedIn, and GitHub need to tell a consistent story that you are already a competent data scientist (because if you have the skills and have done solid projects, you are!). So, instead of writing “Data Analyst seeking Data Scientist role,” you might say “Data professional specializing in predictive analytics and machine learning.”

When it comes to your GitHub, make sure to put your best stuff at the top here. This is especially important for analysts, since your coding skills will be under more scrutiny. So,

Pin your best ML projects at the top

Write clear READMEs that explain your approach

Make sure your code is well structured and documented, showing you understand software engineering principles

And add visualizations and results to showcase the impact, which should be easy for you with your background!

Once it’s time to apply, prioritize hybrid roles. These are positions that sit between traditional analytics and data science, and they’re often an excellent stepping stone.

For example, lots of companies (including big tech firms like Meta and Amazon) have roles that they call “Data Scientist” but are actually more like advanced analytics positions. And honestly at many companies, the lines are blurry anyway. Use this ambiguity to your advantage!

When you’re networking and preparing for interviews, leverage your analytics background. Use your deep understanding of business context, clear communication skills, and examples of how you’ve influenced the business to deliver measurable impact. Other candidates who may be more technical than you might struggle with the business and communication side of things. So don’t be afraid to lean into your strengths.

Remember, this transition isn’t going to happen overnight, and that’s okay. What matters is consistent progress. Every line of code you write, every concept you learn, every project you complete — it all adds up.

If you’re feeling like you need some support with your data science/ML career, here are some ways I can help:

If you’re interested in becoming a Data Scientist, I have a really comprehensive video on YouTube with a complete roadmap on how to go from an absolute beginner to your first job.

Or, I have a free 80+ page e-book with the same roadmap info, plus links to learning resources, checklists, and more!

If you’d like to chat 1:1, you can book a call with me here.

Note: This post contains affiliate links. If you make a purchase I’ll earn a small commission, at no cost to you. Thank you for your support

The post How to Switch from Data Analyst to Data Scientist appeared first on Towards Data Science.