Honestly, it’s been life changing for me.
That drew me into data science and started my career in this field for over 5 years. There, as both a data scientist and a machine learning engineer, I’ve secured over $100,000 in offers from large technology companies to small startups.
But looking back, I made so many mistakes that I actually wish I had a clear roadmap to go from complete beginner to expert.
In this article, I want to detail the exact roadmap you should follow if you want to relearn Python for data science.
Let’s get started!
Is it worth learning Python?
Is it worth learning Python in the age of AI?
AI is extremely powerful and tools like Claude Code can literally do everything for you, but that doesn’t mean learning to code is a waste. In fact, its value is increasing.
Let me just say this directly: this “atmosphere code” is middling at best, and extremely error-prone, which is ridiculous.
Will AI generate poetry? Will it be as good as Shakespeare’s sonnets?
Probably not.
The same analogy applies to AI-generated code. People see a working solution and assume it’s perfect.
In fact, being able to understand and read code properly is becoming a superpower these days. You can instantly determine where the problem is and debug it without wasting time “prompting” the AI to fix the problem.
Finally, if you want to become a data scientist, you need to be able to pass a coding interview. And unfortunately, the use of AI is not allowed.
environment
To actually run Python code, you first need something called a “development environment.”
These environments essentially help you code by providing syntax highlighting, indentation, and general formatting.
For complete beginners, we recommend a notebook environment such as:
- Google collaboration — Completely online, no need to download anything locally.
- Jupyter Notebook / Anaconda— It provides an all-in-one download solution for Python and major data science libraries.
You can also download the integrated development environment. This is often used for writing professional/production code. My two main recommendations are: PyCharm or VS code. You don’t have to worry about which one you choose as both are equally good.
One thing you may be wondering about is an AI coding IDE. These are incredibly powerful and the most common ones I recommend are: cursorand Claude .
However, given the efforts we are making; learn We don’t recommend using the AI editor to write code in Python. That doesn’t make sense.
basic
Once you have your environment up and running, you need to learn the basics.
This will probably be the most difficult part of the journey as you will literally be going from zero to one.
If it’s difficult, that’s completely normal.
Every successful data scientist or machine learning professional has been in the exact same situation, seen the results, and endured it long enough to build a career they love.
The main areas you need to learn are:
- Variables and data types
- Boolean and comparison operators
- Control flow and conditional statements
- For Loops and While Loops
- function
- Native data types (lists, dictionaries, tuples, etc.)
- class
- package
data science package
Once you understand the basics, it’s time to focus on skills specific to data science. That’s where the learning goal is.
Start by learning some of the more specific data science packages. Here’s what I recommend:
- Numpy— This is for working with vectors and matrices, which is what most machine learning is built on.
- panda — This is for data frame manipulation and analysis. You need to learn data science because it’s in the name “data” science.
- matte plot rib — I can’t tell you how many assumptions I made about the data in order to visualize it and make it happen.
- Sci-Kit Learn— Python’s leading machine learning and statistical learning package. It’s easy to use and a great entry point to machine learning.
We don’t worry about learning deep learning frameworks such as: TensorFlow, pie torch,or jacks At this stage. This will come a little later, but is not required for many entry-level data science positions.
project
If there’s one secret to learning Python quickly, it’s by doing projects.
Projects require you to find solutions, unblock yourself, and be creative when it comes to programming.
There are many ways to build ML models from scratch or through courses, like Kaggle.
However, the best projects are the ones that are personal to you.
These projects are intrinsically motivated and unique by definition. So when it comes to interviews, it’s actually interesting to discuss because the interviewer has never experienced it before.
Here’s a basic guide to coming up with ideas for your project.
- Please list five areas of interest outside of work.
- For each of these five areas, think of five different questions that you need answers to and that you can write a Python program to solve.
- Pick the one that excites you the most and start implementing it.
This process will only take at most an hour.
So stop Googling people like me for projects and look internally for what you should build. Because it’s much better.
Remember, we’re not looking for perfection or building a rock star portfolio. This is all a learning exercise.
advanced skills
After completing a few projects, you should have a fairly basic level of data science Python skills.
Now is the time to level up and start learning more advanced Python and software development skills.
These are the core areas we need to study:
- Git/GitHub— This is the gold standard tool for code version control.
- PyEnv— Learn how to effectively manage local Python versions for different projects.
- package manager— Being able to manage libraries and their versions is important for software development, so you need to understand tools such as: pip, poemand ultraviolet lightis essential.
- circle CI— This allows you to efficiently and continuously test and deploy your code, speeding up your development process and enabling you to migrate quickly and with confidence.
- self-made— Macs don’t natively ship with a good package manager like apt on Linux machines. Homebrew is a solution to this problem and has been dubbed “MacOS’s missing package manager.”
- AWS— Cloud storage and model deployment, and many other uses.
- Advanced Python— To upgrade your Python skills, you should start learning more advanced topics such as generators, decorators, abstract classes, and lambda functions.
This basic technology stack is what I used at every company I worked for as a professional data scientist and machine learning engineer.
Data structures and algorithms
Unfortunately, all the Python skills you’ve learned won’t always help you get hired.
The coding interview process is a bit complex in that you are often asked to answer coding questions related to data structures and algorithms (DSA), an area that you rarely use on a daily basis as a professional data scientist.
How much you need to study DSA will depend on the specific data science role you’re looking to get into.
If you’re looking to get into more of a machine learning role, you’re much more likely to face questions in a DSA interview than if you’re looking to get into a more product or analytical data science role.
In any case, DSA is a necessary evil these days and you need to invest some time in it if you want to get hired.
The biggest cheat code I’ve found is that not all DSA questions are created equal. In reality, only certain topics will come up in the interview, such as:
- arrays and hashes
- two pointers
- double sliding window
- linked list
- binary search
- stack
- trees
- heap/priority queue
- graph
Start learning dynamic programming, experimentation, and bit manipulation to avoid shiny object syndrome.
The above topics will give you the highest return on investment. Everything else is noise and has no value at all.
It’s very easy once you put it into practice. It is recommended to take Neetcode’s DSA Course and, Leetcode’s Blind 75 Question Setis the most commonly asked question in interviews.
The shortcut to improving your DSA is to work on it every day for 8 weeks. That produces results.
parting advice
Frankly, there are no secrets or hacks to mastering Python.
The real secret is consistent practice over a sustained period of time.
When I was learning Python, I was coding for almost an hour a day for three months. This was a lot of coding, and don’t get me wrong, it was a lot of effort.
You will have to spend many hours, but it will work out in the end. You need to give it some time.
Coding changed my life and gave me a career that I love and have seen myself working for decades.
This short investment of time and energy paid off far more than I could have imagined.
If you’re reading this and want to start your journey of learning Python to become a data scientist, that’s great.
However, Python alone will not be adopted. There are several other areas you need to learn to secure a full-time job.
So I recommend this articleHere’s a breakdown of everything you need to study to land your dream data science job.
See you there!
One more thing!
Join our free newsletter. Every week I share tips, insights, and advice from my experience as a data scientist and machine learning engineer. Additionally, as a subscriber, my Free resume template!
dishing the data
Weekly emails to help you land your first job in data science or machine learningnewsletter.egorhowell.com
