So my normal way of learning a new language or computer tool is pretty simple and probably the way most people do it: Dive in.
There’s a problem that lead to that language, google to find & install the tools, and then google to solve the problem. Repeat until the solution is good enough.
It can be slow but rewarding and it works. The price is certainly right.
The upside is that you never have to learn the tedious rules of the ‘proper’ way to do things, and you mostly learn the theory by doing. The downside is that it sometimes leaves me wondering, when I’ve solved a problem in some awkward way, “How would a properly trained person do this?”
Also because each of us has a blindspot and is enormously stupid in our own particular way, there will be some epic stupid programmer tricks. Yes, frustration is the pain you feel just before you learn something but no one really likes pain.
So when I decided the forecasting & other predictive stuff we’ve been needing to up our game on needed some attention I decided to use Python. Not knowing Python, I was tempted to just jump in but while I was reading a post by Lisa A. Chalaguine and saw her recommendation for the Coursera course Python for Data Science, AI, and Machine Learning Sounds perfect!
So I took the course. They say it’s a 5-week course assuming you finish one module a week. I finished in a bit over 24 hours elapsed time, maybe 7-hours of course time.
The title of the class is a bit aspirational. Yes, I did learn some Python but it was more learning about what was possible than actually how to do things. It was like watching a half-hour segment on HGTV on how to install a swimming pool. Sure, they mention all the big steps (decide where to dig, dig, put in liner, hook up plumbing, pave deck, fill and swim!) but you know you’re not really getting the full dish.
Well that’s ok. When I was signing up for the course they asked what certificate or degree I as pursuing. I’ve got two degrees, and I’m not interested to pay for more, so I decided on a certificate and on the list of courses was a more useful sounding Data Analysis with Python. This was much more useful. Covers all the major aspects – data cleaning, regression, multiple regression, testing, and more importantly the automation of the model.
A bit of hands on to solidify things in a final project module, and already I feel like I’ve got a reasonable foundation to start from.
Will I continue with the other three courses needed for the certification? Well, the 7-day trial ends in three days and I think I can make it.