Principle Component Analysis and Why Data Science Scares Me

(and why it shouldn’t)

So, what are you about to read?

• One (brief) example of how data science is wonderfully clever
• One (less brief) example of how overwhelming it can be, no matter what your background is
• Tips and tricks for getting past the anxiety

If you’ve read that bulleted list up there, you may have already noticed that I may have mislead you with the title. Am I going to talk about principle component analysis (henceforth PCA)? Yes…but only sort of. I first heard about PCA a few months ago, when I saw a really neat article, basically just breaking it down. It started off simply, and it was fun to try to understand. The author presents some data, and explains that sometimes when you have lots of dimensions that describe your data, it can be clunky, slow or otherwise inadvisable to work with; you want to reduce the number of dimensions you look at. Do you just delete a column and move on with your life? As easy as that would be, you may be getting rid of some valuable nuance in your data! So what do we do? Cue PCA! This article has me hook, line, and sinker, straight out of the gate.

As I’m reading this article my internal mathematician (the one who got a BS in Mathematics only a couple of months before) treated it like a test — am I still “good” at math? Do I still understand all the notation and concepts I had lived and breathed for four years? The long and short of it was, nope. Hard no. Absolutely not. My heart sank as I read further and further, and the test became more and more real. It became a test of my intelligence, and I had failed miserably. That tiny voice in my head saying “you should try data science, it seems fascinating!” was silenced by a fear of failure.

What I’m here to talk about is that feeling that data science is too much, too hard, to math-y, and just plain too scary. You might think it seems So! Darn! Cool! and yet maybe you can’t bring yourself to actually put the time, effort, and emotional investment you need to succeed. After all, why bother in trying for something you can’t actually succeed in? When that feeling came up for me as I debated studying data science, it wasn’t unfamiliar. I had felt this way nearly every single day of my math degree.

Regardless of your educational background, mathematical or not, pursuing data science is overwhelming. There are so many different fields it pulls from, and they’re all complex and intertwined. But isn’t that what makes it seem so interesting? All of those fields, blending into one domain that takes a whole mess of data and actually makes it mean something! I have often deprived myself of learning opportunities because I convince myself that I’m not smart enough to do well. But there are other times where I’ve overcome those feelings, and I hope that by sharing the strategies I’ve employed, maybe I can help you do the same.

In no particular order, here are my top ways* I work to empower myself to pursue big scary goals (like learning data science):

1. Acknowledge your victories: Learn to recognize and make a big deal out of all the times you have succeeded on the way to succeeding in a larger goal. Sometimes you’ll be really stuck. You’ll feel like packing up and going home. I felt that bi-monthly in my undergrad, and it was tough. I wasn’t particularly well versed in thinking back to past successes, but I wish I was. Now that I’m in a formal learning environment again, I occasionally get those feelings of oh god, what a mistake, I’m not understanding this (or, just as commonly, everyone else is getting it and clearly I’m the dumbest one in the room). The only way to counteract this genuinely, in my experience, is to focus on what I have done well. Did you learn how to write a function in Python? Did you figure out the purpose of an abstract class? Congratulate yourself! Write it down! Tell someone else! These smaller victories mean absolutely everything — you can’t achieve larger success without them, and most of the time larger success is just a bunch of these smaller victories strung together. I’ll be honest, sometimes it feels awkward and self-congratulatory (because, well, it is), but thinking you’re not smart enough comes from ignoring your own successes and downplaying them because they fit the narrative of not being good enough.
2. Find a supportive group of people: Community, as I define it, is a group of people who share values and goals — and it’s one of my most treasured values. Your community in data science is hopefully full of people who want to see you succeed and are willing to help you along the way. Sometimes that means checking your code or your math for errors, or explaining a concept when you’re confused, but more importantly, if you’re feeling stressed, they notice and find the time to help get you to a place where you feel more comfortable and confident. Rely on these people, and be a part of their community. It just feels good to help and be helped, as corny as it sounds.
3. Ask questions (in front of everyone): When something doesn’t make sense, ask people about it. And don’t just quietly ask the person next to you (although sometimes that’s helpful), but ask in front of the whole class. Getting used to not understanding something and owning the fact that you don’t get it can help you shift into a growth mindset. Getting past feelings of confusion won’t happen by maintaining that you’re confused, but by recognizing it, and then articulating what is confusing you. This is especially true when you have to ask your question in front of people you respect — this forces you to boil it down to exactly what you don’t understand, rather than just throwing your hands up and saying well I just don’t understand any of it! No one understands it all, but the people who understand the most are the people who’ve delved deep into the particulars of what they don’t understand.
4. Take a breath and a step back: Sometimes you’ll start feeling that feeling, the one you know is going to take you to a bad place, where you can’t do anything but stress about all you don’t know, all the catching up you have to do, how impossible it all seems, and so on. When that happens, learning isn’t going to happen. You might just convince yourself that learning is never going to happen, and decide to quit. As soon as you catch wind of that feeling, I encourage you to close your computer. Just close it. It sounds ridiculous — you’re trying to understand some hard concept, but in order to do so, you have to stop trying to understand it? — but honestly, your brain has long since left the learning station. If you’re panicking about how much you don’t understand, you’re not in the right mental space to keep pushing yourself. Go get some water, go for a walk, find someone to talk to (remember what we said about community?) — do anything but work on what plunged you into that emotional state. Then, when you’ve had a minute and are feeling a little better, you can come back. Or, if you’re still not there, try tackling something else, build your confidence with that, and then you may feel more ready. Taking a momentary step back has never hindered anyone’s learning!

To be honest, I hope that all sounded somewhat obvious — I know it did to me as I was writing about it. But when you’re feeling nervous about whatever your big goal is, it’s hard to remember all the ideas that were so obvious before panic set in. So I hope this helps next time you’re feeling nervous; I know I’ll look back to it soon as I try and tackle PCA for real.

*There are other helpful techniques out there that I haven’t included/thought of — please let me know if you use anything I haven’t listed!