Why and how I started learning Python and Machine Learning pt.2
It has been over a year since I decided to document my python/ML/Data Science learning path online a bit. A lot has happened in the meantime, especially in my personal life - which is why I was less active online in 2018.
In this part I will briefly go a bit deeper into what online resources I use to continually learn and apply my knowledge of Python, ML and Data Science in order to grow professionally.
I try to listen to podcasts as much as possible during driving, cleaning, walking and other activities that do not require my full concentration. I don’t have a definitive list of podcasts I listen to, but I rather pick specific episodes from different podcasts that catch my attention. Just search for “data science podcasts”, “machine learning podcasts” or “python podcasts” - there is a ton of them. Listening to these has been very useful for me to get to know the respective ecosystems, find out about popular libraries and techniques used.
Furthermore, they help with motivation and provide inspiration as they often feature beginners, or people like me, who transition from different backgrounds into python/ML/DS.
Some of the podcasts I listen to include:
- Talk Python To Me
- Data Skeptic
- Super Data Science
- and more…
Twitter is a strange place, one can form it into almost any kind of social place they wish - so I use it primarily for learning python/ML/DS. It acts as a content curator for me, I get interesting articles, cheatsheets and tutorials from it. Sometimes I just read them, sometimes I act on them and try to replicate their content or use them as inspiration.
I also use LinkedIn in a similar fashion, but less frequently and a bit more randomly, but it can also be a source of similarly interesting content.
See my Twitter profile for a current list of who I follow.
edit: I added books and blog sections shortly after publishing this article:
I always have 2-3 books that I read at a time - and at least one of them is related to my profession. I am not much of a “learn by reading books” guy, so I utilize books in a different way - they serve me as a quick overview of something, or a detailed source for something very specific when needed. I often buy books in bundles, like Python or Big Data bundles from Packt on Humble Bundle, which give me way too much to read - so I try to speed read through them, or simply skim them to get the most important ideas. This approach allows to maximize benefit with minimal effort - while still having the books at hand when more detailed info is needed.
This is an “opposite direction resource” - writing this blog is helping a ton to learn - I think everyone knows the “if you can’t explain it simply you don’t understand it” principle - it definitely applies here, I have to make sure I know what I write about, which forces me into deeper understanding.
Same principle applies to contributing to StackOverflow, but apart from an occasional answer or vote here and there I don’t participate in that community much for various reasons.
All the listening and reading would be useless without practice - and that is where Kaggle comes in. I only started using it actively very recently, and I already regret not doing so much earlier. Solving real-world(ish) challenges to get experience in ML is the best way to learn, as you get practice, feedback and discussion with others. I am looking forward to getting more out of this platform in the future.
All this is a bit vague, but all these sources combined have been a great help to keep the learning/exploring motivation up. Here is a short list of examples:
- I learned about the existence of Dask because of Talk Python To Me podcast - and it’s on my “try and blog about” shortlist now.
- I got to understand how important (historical) context and qualitative analysis is during the famous Titanic Kaggle challenge
- I learned how to deploy ML models to a website when trying to publish my model for the same Titanic Kaggle challenge
- I get constantly reminded that ethics is a thing to keep in mind
- I got hold of great collection of DS cheatsheets which will surely decorate my wall when I land my first DS job
- Quickly skimming through a Django book gave me an overview of the framework and it’s general principles.
- Writing about building a simple classifier prediction model made me aware about a potential decrease in prediction accuracy when using one-hot-encoding technique for categorical data.
Do you have some useful resources you use? Please share them in the comments!
As always, keep learning and having fun!
In case you missed it: part 1 of the series.