Hello! Machine learning sparked my interest, and I’m ready to dive in. I have some previous programming knowledge but I basically start at zero in data science. So naturally, I don’t really know where to begin this journey. I’ve researched for resources and roadmaps to learn machine learning and created my own basic roadmap just to get started.
Math - 107 hours
Programming - 135 hours
Machine Learning - 200+ hours
Please give comments on it and or advice on better/more efficient ways to learn. Thanks!
4 Likes
I’ve been diving into data science, math, and machine learning for about a year now, putting in around 500-1000 hours of study. It’s not just about learning math and algorithms; you also have to pick up a bunch of new skills. If you know some basic Python (like numpy and pandas), you can skip the computer science basics for now.
Start with Andrew Ng’s courses—they’re great for getting a solid overview. Don’t expect to become an ML engineer right away; it takes practice and projects to really understand. The first course uses Octave, which is a bit odd, but it helps with understanding linear algebra.
Math takes time, especially linear algebra. If you really want to dig deep, consider getting a proofs book and a Chegg subscription for help with exercises. It’s really important to apply what you learn and build related skills.
Here’s a plan to follow:
- Master pandas (check out Harrison’s book).
- Learn SQL (try DeBarros’ book and practice on CodeSignal).
- Get comfortable with regular expressions (regex101.com is helpful).
- Read a book on data visualization.
- Learn matplotlib—it’s tough but worth it; I remade graphs from “Better Data Visualization,” and it was a challenge!
- Sign up for AWS and Google Cloud to understand their services.
- Listen to data science and ML podcasts.
Your goals matter a lot. I’ve been at this for a year and learned a ton, but there’s still a long way to go. I estimate it’ll take about 3-5 years and around 2500 hours to feel competent—meaning I can develop, deploy, and monitor different models.
Balancing study with being a resident physician has been tricky, but I usually manage 6 hours on days off and 2-4 hours on workdays. I read for 2 hours, work on math for 2 hours, and study ML for another 2 hours. When work projects come up, I dive into them fully and apply what I learn to my own datasets too.
Eight-hour study days didn’t work for me, so I suggest starting slow and gradually increasing your study time if you enjoy it. It’s definitely doable alongside a busy life if you make it a priority!
1 Like
Just curious. You went from nursing (RN?) to MD, and now you’re learning data science, math, machine learning. Is that mostly a hobby or are you planning to pivot to more data-centered roles (research, industry, etc)?
2 Likes
I have a background as a registered nurse and then got my BSc in biochemistry before moving on to medical school. In my clinical practice, I see a lot of opportunities for machine learning to help healthcare providers make better decisions for patients. The problem is that healthcare data is mostly untapped due to HIPAA regulations. Those who know data science often don’t have access to the data, and those who have access usually don’t know data science. There are some exceptions, but that’s the general situation. I’m trying to help fix this gap.
I also really enjoy the learning process. It’s like a hobby for me, but I plan to use these skills in my work too. Honestly, I love medicine, so I guess you could say I just really enjoy my work!
1 Like
Hey mate I don’t have any past experience and am really interested in AI and all this stuff and I’m 17. Can u help me out like what step I should consider taking so it will help me and your own any word that u think someone would have said you before starting this journey. Hope the best. Hope you will reply thanks
1 Like
Okay sounds good but I have to modify a Lil bit as I don’t even know programming 
so I have to learn python in the way. So all the best G
1 Like
Thanks! I like your approach too. Here’s a different method that I think could work well for you and others.
Start by checking out the Python section on W3Schools and go through the first 36 chapters. If there’s anything you don’t get, you can look at books like “Python Crash Course” or watch YouTube channels like Navin Reddy (Telusko) and Codebasics.
Next, search for the top 10 machine learning algorithms. Pick one algorithm at a time to implement. For example, if you choose “linear regression,” you could check out resources like:
a. W3Schools: Linear Regression
d. GitHub: Multivariate Linear Regression
Keep following this pattern for all 10 algorithms. By the time you’re done, you’ll have learned a ton about pandas, sklearn, matplotlib, seaborn, and other tools you need.
I believe this way, your reading will stick with you better!
Start from high level → then go deeper
select a topic that u are interested in, right away try to train models
Learn by developing
Train, validate, evaluate on test set
otherwise there is possibility u may give up on the way … because so many low level subjects to learn
1 Like