It’s been a while since I’ve talked about my data analysis self-study.
I’ve been trying this and that, but haven’t felt anything was worth writing about. I mean, who would want to know that I tried something and failed, right?
Oh wait. Me. I would want to know.
When I’m about to try something new, like skincare or a restaurant, I look up blogs for reviews. I try to see if I can relate to the blogger and put myself in their shoes–Would I have failed as well?
It saves me a lot of effort because someone else has already gone through the experience for me.
That’s why I’m writing about all my data science-related updates so far, incomplete and disorganized as they are. Maybe it’ll help.
As I mentioned in a previous post, I paused on the Microsoft Professional Program in Data Science for now. I struggled with the lack of hands-on exercises. I do think I’ll continue it again in the future as a Microsoft certificate is pretty weighty on the resume.
Around this time I started questioning whether Data Science was really for me. I had an inkling that its the visualization aspect I’m more fascinated in rather than the science and analytics itself.
Discouraged by the Microsoft program’s statistics course, I dived straight into coding with Interactive Python. I highly recommend it to people who are learning to code from scratch.
Not for me though. Too much reading, not enough doing.
Speaking of “doing”, I then completed Data Quest’s Python Basics. Data Quest is an advocate of the ‘learning by doing’ approach so it was quite the antithesis to Microsoft’s lecture-rich MOOCs. At the time, I considered DataQuest the best match for me.
The rest of the courses require you to sign up for a paid account, something I wasn’t sure I was willing to invest in given that:
- While I could “do” stuff, I didn’t understand why I did stuff. DataQuest was very good at hows but not whys.
- Even a beginner like myself could tell that some of the code examples could be simplified further. I felt it wasn’t good to be teaching students to code in unnecessarily lengthy syntax.
I started reading Storytelling with Data over the holidays and fell in love.
Its what I am passionate about: Being able to tell stories with data. Yes, I have to extract, clean up, merge, analyze, predict, etc etc the data… but its the end result I get excited about.
It gave me a better idea of what I really wanted to learn. It wasn’t so much about data visualizations, but rather data communication.
Inspired by this, I looked up how to make data visualizations specifically rather than the whole of data science. The search led me to FreeCodeCamp who offered a data visualization certificate.
I liked FCC. Their approach was a nice marriage between Interactive Python’s explanations and Data Quest’s learn-by-doing approach. A bit too much hand-holding, but the projects made up for it.
Zoning in on data viz I learned about Udacity’s data visualization course. It required a lot of prerequisites however so I thought it over for a while.
Then my day job picked up and I no longer had the luxury of time to self-study. I used whatever spare time I could get to read about data storytelling, and it was from there I found out that a thing like data journalism exists.
Assessing my options, it’s probably closer and more realistic to what I’m aiming for rather than data science.
Another thing I picked up from all that reading: There are people like me, educational background and all, self-studying data science. And these people (well, person) recommended Udacity’s Data Analyst Nanodegree.
I tried it out, starting with one of Udacity’s core courses: CS101.
I loved it.
Even though it covered almost exactly the same thing as DataQuest’s Python Basics. I understood Udacity’s version so much more. And the code? Elegant.
So… what now?
Well, I’m a bit worried about Udacity. Like edX or Coursera, each course is taught by a different teacher and it might be that I’m enjoying the CS subjects right now because of how good the professor is. I can’t say it will be the same for the rest of the courses.
Also, it’s not cheap. Like other MOOCs, you can audit the lesson videos for free. But unlike other MOOCs, to qualify for the nanodegree, you have to sign up for their projects-based curriculum at a steep $200 a month (50% refund if you finish within a year).
There’s also this data science career guide which talks through the best MOOC sources for each data science skill. You could say, take Python classes from Coursera, a Statistics course from Udemy, etc.
Assuming the guide and I have similar preferences, this would be the optimal route.
What concerns me is that the lack of a consistent provider means I won’t have a portfolio at the end of it (unlike DataQuest or Udacity) nor a certificate (Microsoft).
Oh, and that data journalism thing? That might also be worth exploring.
Codecademy has a very light touch; it’s a good way to get a feel for a programming language if you’ve never tried it before. If you’re actively trying to learn the language though, I’d suggest looking elsewhere.