Please tell me how to upvote this 100 times!
Finding likeminded people like you is what makes internet so exciting…
I read your article: Use Kaggle to start (and guide) your ML/ Data Science journey — Why and How
and it was so inspiring and so aligning with my experience, that I wanted to explore more about your writing and found this 2nd gem. Almost every word in these two articles are ringing so true, that I highlighted almost all paragraphs in my diigo highlighter.
Unfortunately, I came to the very same conclusion of yours, after so many years of doing learning through courses and books in the wrong way. In the last year or so, when I reflected back to see how I was doing, I realized that I have truly learned ONLY in the few big projects that I did, which then I found them to be very meaningful and useful to me and for others. The target was so compelling that made me learning everything just for the sake of reaching the goal of developing that Android app or that electronics project that I dreamed of. Everything for that sake was flowing automatically. I was addicted to solve struggles for weeks. And when I have finished the project, it was giving so much joy whenever I use the app or somebody used it. The side effect was learning to code Android or designing electronics and so on.
I have even tried to learn Android programming before that project onlyfor the sake of learning, and I have failed to take it seriously. It lacked the true motivation to dig deeper. My learning was shallow. And after 1–2 years I have lost my investment in the time to learn the course and forgot almost everything. But with doing a meaningful project, the learning was so deep and it changed me forever.
I am sad that I have not discovered this years before.
My current understanding on how to effectively learn is:
- When you have a meaningful project idea, go for it by all means. Don’t let anything to make you lose this important opportunity to learn. This is the only way to make a progress in learning. Nothing else will come even close. The grit that you will get because it is meaningful to you makes you able pass the roadblocks and all inevitable struggling. No books, no courses, no college… Nothing but meaningful projects. For data science and machine learning, you mentioned in the kaggle article, that for this particular topic, there is one thing makes it difficult to apply a project that you find meaningful. That’s because in machine learning getting a good dataset is always very difficult. And kaggle is the way to go to solve this issue.
- Online courses and tutorials are very important indeed. When you pick a project, you need to learn a lot about every gap in your knowledge about that topic. Try to choose top-down approach of courses/books/tutorials. I have finished fast AI course and it is a good example of this top down approach of teaching. By showing the student the usefulness and how to do a state of the art project at the very first few lectures (top) and then with later lectures going deeper (down), it will give you the sense of accomplishment from the very beginning. It will give you confidence, and you will learn how this is an easy thing to be done. I have learned that I have to stick with courses, papers and books that have solved real projects as jupyter notebooks or android projects that I can find in Github of the author. Because I think the best way is to reproduce the author’s example by simply clone it and run it, and then after you are confident that you can make this working indeed, you tweak here or there or modify it to your desired project. This top down approach is truly lacking in most courses and books.
The details of why this approach is better and all the insight that you have, I got it in a painful way in the past years. And I can reassure you that I am thinking exactly like what you’ve said. I have never found a clearly written, and thorough explanations like yours to confirm my believe. Those 2 articles you’ve written made me happy that I am not alone.
Perhaps, we who are trying to catch up with experts are understanding each other better, due to the Curse of knowledge . And that’s why Jeremy Howard perhaps is much better to teach than many other experts, because he was a hobbyist, and knows how things are working in teaching and learning. In fact I have even heard from him, that he is still dedicating half of his time to learn from online courses and tutorials. He is truly and inspiring person.
I read all your twitter feed, and read in one of your tweets that you wanted our opinion on this article since it means a lot for you and I can confirm that I have got the very same experience like yours. SO I thought you would be interesting in details which do not fit in few tweets.
Wishing all the best in your endeavors and learning!
