Data Science


I just finished reading the book ultra learning by Scott Young and I thought that this concept could really help many of you on our data science strands Scott used this approach to learn all of the MIT computer science coursework in a single year which usually should take four years of learning first what is ultra learning according to Scott ultra learning is a strategy for acquiring skills that is both self-directed and intense I think that for many of the followers of this channel I expect that you are self-directing your data science learning journey even if you’re not if you’re going about a more traditional approach to learning data science such as a university coursework a certificate etc …

I still think that this article could really be useful to you in the book Scott has nine principles of ultra learning and in this article I’ll highlight a few of the most relevant ones for your study of data science if you want to learn about the rest I highly recommend reading this book and it’s linked in the description below this article goes very nicely with other article about these stages of learning data science which I’ve also linked above and below I recommend watching that one after this if you haven’t already okay let’s jump into the first principle of ultra learning which is called meta learning this is what sets the groundwork for your learning journey you should take the time to get a broad understanding of the field in this case it’s data science and create a learning roadmap.

Scott suggests that you should actually take about 10 percent of your total study time to build this out first you should think about what exactly you want to achieve by going on this data science learning journey do you want to be able to build a specific app using data science or would you like to learn enough data science to be able to get a job for example or is there some other reason all together here when you’re thinking about this the more specific you can be the better next you should start to map a learning path to get to your specific destination.

I look at this very similarly to building out your own curriculum for a self-study course at the foundation data science is really just a combination of programming and math and I’d reference my articles learn to program for data science and how to learn math for data science for more information on how to build out these legs of your roadmap they’re linked above and below I’d also recommend looking online for data science course curriculums most universities post these and you can see what classes or concepts the universities are recommending for new data science students to take from this you can figure out.

Exactly what courses you need to attain your specific goal after you have a learning path you need to be able to sit down and do this work this brings us to the next principle which is focus this isn’t necessarily specific to data science but it’s a very valuable skill to learn the keys here are scheduling time to actually do your work staying committed to it and limiting distractions I have another article which I’ve also linked above on how to stay productive and motivated when learning data science that takes a significantly deeper dive into these concepts so the third principle of ultra learning is directness and this is my favorite one by far it suggests that you should just dive in and start doing the actual activity that you’re trying to learn when you get your hands dirty with a real data science project.

You almost immediately understand your shortcomings I constantly recommend projects and I maintain that this is the most efficient way to learn data science when starting out on a new project I usually work until I get stuck on something and I open a few tabs I learn about the area that I was struggling with and I figure out how to apply it back to the project I was working on when you take this approach it can be unbelievably frustrating and infuriating but it assures that you’re learning exactly what you need to complete the task at hand I recommend leaning into this frustration and beginning to enjoy it because most of the data science follows this process in this pattern.

This image has an empty alt attribute; its file name is source-4280758_1920-1024x683.jpg

Even at a high level you know if this isn’t something you’re comfortable with if you don’t like that frustrated feeling data science might not be for you or you might have to learn to love that feeling if you want to learn more about the actual fourth and fifth principle which are drilling and retrieval again feel free to check out the book that I’ve linked below okay back to feedback this can be really scary and uncomfortable but for the learning process it’s extremely important data science is nice because there are many clear feedback sources.

First with programming most times that you do something wrong you’ll get an error you should be able to read these errors carefully because they generally give you a really good hint about what is wrong I also recommend using Stack Overflow if you can’t figure it out just from the error message you can usually just copy and paste the error message into Google and I guarantee that someone else has faced a you know the exact same problem or something very similar before another way to get valuable feedback is to ask for it via social platforms kaggle among others is great for the specific use case.

I also use reddit I think that there’s some great forms out there you can also ask other learners or data scientists that you know to give you feedback on your projects for example if you’re looking for a job this could be a great way to open an informational interview you could say that you’re trying to improve your data science skills and wonder what someone in the field thought about your particular analysis again this could be scary but it’s the quickest way to learn generally.

Actually don’t mind giving feedback to you guys but I’m getting a huge number of requests lately so please don’t be offended if I don’t get to yours or that it takes me a really long time if you want me to review your work please make sure the necessary steps to improve yours it absolutely drives me up the wall to see errors in resumes and projects that they’ve gone incorrect it I’ve made a playlist with all of the relevant articles for resumes and projects and I’ve linked it below so please watch that before reaching out asking for my help reviewing these things I’ve also been thinking about doing a live stream where I’d review people’s resumes or projects.

If this is something that you’d like to be a part of or you think would be interesting please let me know in the comment section below I really love the engagement downer okay back to ultra learning so we’re going to skip to the ninth and final principle which is experimentation this is how you actually get better at the data science craft in data science there are many ways to solve the same problem and in your earlier learning you probably only learned one approach.

This one way might not have been the most efficient or the most practical way of doing things with experimentation you actually have to forget some of the previous approaches that you’ve taken and you have to come at problems from a different angle I believe that this is what separates good data scientists from great data scientists tinker and learn rather than just using the out-of-the-box methods that they’re previously familiar with experimenting also helps you to stretch your data science muscles and approach a higher level of learning hopefully this article helped you understand this ultra learning approach to self-study.

I really wish I could go back and learn data science from scratch with these concepts in mind again I recommend reading this book if you want to learn more.

Leave a Comment