Zero to Mastery’s Machine Learning & Data Science BootCamp — The best course?
Back when I wanted to start learning about AI and data science, I had no clue which course I should take. There were so many roadmaps, so many subjects differently prioritized by different people…I was lost.
Then, I came across Daniel Bourke’s article about how he created his own master's degree. I saw someone who has this strong desire to learn about artificial intelligence and self-study just like I do, that’s why I could relate to him (or at least him 5 years ago, as he’s now a professional machine learning engineer). Making some research, I found out he made a Data Science and Machine Learning BootCamp based on the knowledge he learned by himself.
As I was touched by his story, I considered giving it a try. And one year after finishing this course and stepping back through the projects and other courses I’ve taken, I’m taking stock of this course.
Structure of the course
This course is made in such a way no prior experience is required. Indeed, here are the different chapters:
- Machine Learning 101
In this chapter, they give you an intuition to understand why we need machine learning (ML) and data science (DS), and they explain how it works from a very high level.
- ML and DS Framework
Here, they walk you through a basic framework you can use in projects (all the way from analyzing your data to optimizing your ML model) without any code, just pure theory
- Learn Python
8 hours of hands-on learning to understand how to use this programming language. I’ve honestly skipped this part as I knew how to use Python, but after skimming through I can tell it’s a decent and sufficient introduction to Python to get started with DS.
- Environment setup
If you get into DS, Conda and Jupyter will be your best friends. And this part walks you through the installation and the basics of these programs, even though the instructor will give tips about how to use them during all the course.
- DS libraries ( Pandas, Matplotlib, Numpy)
These libraries are a must, at least at the beginning ( some people use alternatives but they do an excellent job regardless of your current level ). This part can be a bit annoying as you’ll just learn how to use some functions with toy data or datasets without feeling you’re doing much. But believe me, it is worth it: the fun part is right after this one and you’ll thank yourself for having studied these libraries.
- The ML part, Scikit-Learn, and milestone projects
After a very long (but fun!) experimentation with Scikit-Learn — the module allowing you to build ML models —, you’ll work on two Milestone Projects putting it all together to make some Classification and forecasting models. This part is interesting because it makes sense of why you’ve learned all those libraries
- Deep Learning with Tensorflow
A very brief introduction to neural networks and (another) library, Tensorflow. This library is very popular in the field of Deep Learning.
- and some miscellaneous bonuses…
Bonus tips, resources to learn mathematics, introduction to data engineering….
Content of the course
Because yes, the structure of the course may look great, but it doesn’t tell us about the content. But instead of writing a big chunk of text to tell you what I think about it, I’m going to list all the pros and cons so you can better figure out what to expect.
Pros
- Daniel (the instructor) is a great person. It may not sound important but it’s essential to keep yourself focused on the course, and he does a great job by being cheerful, being dynamic without ever overdoing.
- The theory is well explained and illustrated with great diagrams.
- Each line of code is entirely explained if there’s something new, without being intrusive. It’s great as it makes sure you don’t blindly write the same code without understanding what you’re doing.
- The instructor is demonstrating the “Google is your friend” proverb. Sometimes, he would make a mistake on purpose and look it up on the internet, so the student learns to use the internet to debug their code. Similarly, if he introduces a new topic, he shows that you can easily search for resources on the internet to learn more about it and dig into a rabbit hole because you’re curious.
- A lot of chapters are provided with homework. This homework is a notebook with instructions so you can get your hands dirty and put into practice what you’ve just learned. This is more helpful than it sounds because you may experience the Dunning-Kruger effect and think you know the library because you just watched the video. Please, don’t skip them like I did and regret it later :’)
Cons
- Honestly, the only con I see is that it doesn’t dig deep into the theory but again, it’s an introductory course so it’s not its purpose.
My opinion
As you may have guessed, I loved this course. It was my first introduction to data science and I was addicted: I went through all of the 45h of lectures within a week, even though I also had class and exams. It’s very different from the standard introduction courses you can find on the internet, but the fact it’s so hands-on-oriented makes it very effective.
Indeed, the content doesn’t go really far, but for an introductory course, it’s more than enough. Plus, the goal is not to make you an expert, but to have a general understanding of the big picture and how all the pieces come together. It gets you ready to work on projects and improve by yourself every part of the framework (for example, you can decide to focus on the analysis a bit more on some projects to improve this skill) as you know what could be improved.
As a final word, even if it’s a paid course, I would say it’s definitely worth it to spend a few bucks to get yourself started in DS and AI. ( Oh, and as it’s a Udemy course, don’t buy it full price, “Sales” are very frequent)
