We were promised emotional roller coasters along the Immersive, and that’s a promise well kept. And I’m not talking about the Winnie the Pooh ride at Disney World. I’m talking about serious stuff like Hollywood Rip Ride Rockit with all the spiking, dropping, twisting and spinning, nonstop!
Week 3 (Regression) was by far my lowest. A big picture understanding of statistics and probability can go a long way here. Without it, picking up small pieces here and there right away can be very frustrating, because you never know where each piece would fit, or how you would fit it even if you knew where. It's like doing a puzzle but not knowing what the final image looks like. This applies to both working the regression models and evaluating the performance of the models.
Week 4 was right up next on the list, and then week 2, week 5 and week… Okay, I think you see where I’m going.
But then it got me thinking. If every week was horrible, maybe I was doing something wrong.
I started to feel a little better around Lab 4.02 where I decided to jump away from the starter code and build my own chain of thoughts. The starter code is nice to begin with, but it can get in the way at some point. You will have to develop your own framework eventually. After all, there won’t be any starter code notebooks prepared for you at work. So it's important to take a step back from time to time and review your current framework to see if or where any new tools can fit. I mean it's easy to get lost in all the shiny fancy tools and the very human unfriendly documentations, but at the end of the day, it's where and when to use these tools that matter the most.
Another thing that helped was writing down a to-do list before working on any project, just like for everything else. For example, you'll always want to look for the data, read it in, clean it a bit, plot and explore, and maybe do some outside research, especially on topics that you don’t know much about. Overtime, you'll feel more comfortable about this routine.
Also, don't let questions build up. Things are moving fast and even tiny questions can throw you off easily if not taken care of soon enough. I was embarrassingly confused about the word “model” for quite some days as it seemed to mean different things. I thought I should sweep it under the rug since everyone else was so worry-free about this word, until one day I couldn’t pretend any more and asked Jeff during class. That was a surprisingly straightforward answer and immediately connected a lot of dots for me.
Out of nowhere that reminded me of a fairytale by Hans Christian Andersen: The Princess and the Pea. In a bizarre way, the story tells how the princess didn't sleep all night and got bruised by a pea placed under 20 feather beds because her skin is so sensitive after so many years of royal life. (Sorry I’m a horrible story teller.)
But it does have a point. I need to become more sensitive to the framework and all the tools, and that requires practice. And the best practice I know is to actually explain things like you're teaching.
So from the next blog post, I will pick something from the class and talk about it like I own it. Maybe it can be of some value. And that’s my pea.
