On Cooking & Data Science
First Course: Learning a New Language
I have recently become a big fan of the Partially Derivative podcast. The hosts are funny and the episodes are easily consumed. A win all the way around (clearly, you should check them out). Back on topic - the latest episode, Invisible Data Science, (released on my birthday, no less) mentioned the use of machine learning to come up with recipes for salads. Yes, salads.
They went on to talk about learning how to cook, which gave me the inspiration for this blog post. I enjoy cooking, and in general, I think I’m a pretty decent cook. However, I never really thought about how I learned to cook. Many times, I shoot from the hip / wing it / get creative. Fortunately that works out more often than not.
Main Course: The Meat
At this point (assuming you’re still reading), you may be asking yourself “How on earth does that relate to data science?” That certainly is a valid question, so I’m going to answer that in a (I hope) humorous, entertaining way.
In the earliest days of chef Charles, I definitely started with simple recipes that only had a few ingredients. Ease into things, or keep the stakes low, if you will. Using this approach, I was able to build my confidence, adding ingredients and tackling incrementally more challenging recipes. Of course, not every attempt ended in success - fortunately, I’m not a picky eater. The times when I would try an adventurous recipe for other people were when I beat myself up the most, if they didn’t turn out exactly as desired. I’m OK messing stuff up if I’m the only one who suffers. I really don’t like letting other people down. That’s just part of my personality, I suppose. Ultimately, the takeaway is to learn from your mistakes and to NEVER give up. Perseverance is something that comes up often in the data science world, and I definitely echo those sentiments.
After a lot of practice, a lot of ups and a few downs, I started to develop my cooking style, and even created a small set of go-to recipes of my own - tools in my tool belt. In order to take advantage of the typically gorgeous weather, I enjoyed grilling and smoking various meats. In some cases, I moved past average and became quite skilled. In fact, I may or may not hold the title “The Grillmaster” in some circles!
Coming back to the Partially Derivative podcast, as I listened on yet another long, traffic-filled drive home, I drew this analogy in my head. Over the past year, I’ve spent a large amount of my “free time” studying, reading, and practicing techniques in machine learning, deep learning and data science. You can find some of my early explorations here. There have been ups and downs, walls, plateaus and even some breakthroughs. As difficult as it is sometimes, what I have to remember is that I didn’t become The Grillmaster overnight, so I can’t put so much pressure on myself to become Data Science Master in an equally short amount of time. Now is when I can tweak, experiment, fine tune, heck - HAVE FUN! The more enjoyable I make the learning process, the more likely I am to stick with it and reach my goals. Right now, my short term goal is to complete the Fast.ai deep learning course. It’s deep, but it seems really well structured, and I’m looking forward to digging in (couldn’t resist the pun!) and exploring everything this course has to offer.Share on Twitter Share on Facebook