We made ittttt! …to week six. Half-way there!

Week 06 Topics:

This weeks topics were Web Scraping, NLP, Clustering, and Time Series.

I’ve done my fair share of scraping (and getting my IP blocked) in my day so that wasn’t particularly new. Learning about MongoDB, one of the more popular NoSQL databases, and how to use it to store data was though.

Mongo is one of those things I’ve kept hearing people rave about but never really had any reason to learn anything more about myself. But now that I’ve gotten my chance to take it for a test drive, I think there’s a good chance I’ll be joining the long list of acolytes. But I’ll make the final verdict after seeing how much weight it carries while doing my capstone project..

NLP day was as cool as expected, but it was kind of sad we couldn’t do anything beyond tokenization, stemming, and tfidf. These are more like the standard text preprocessing steps, and not really at the meaty core of NLP.. then again, we have just a day to go over huge concepts some researchers could spend their entire careers on, so the limited scope is understandable.

The morning NLP lecture at least definitely did a great job letting students know that there’s a whoooooole lot more you can do with NLP power.

Alt

(My copy of Jurafsky and Martin, the NLP Bible)

For clustering, there is ISLR, ‘nuff said.
(Anyone else notice how R books talk a lot more about hierarchical clustering, where as Python books do more so for KMeans??)

Time Series day was on Thursday, the last day before our break… and my mind was already long gone into holiday mode.

Other than hearing that there is something called ARIMA and that R is better for time series work, not a NaN thing was learned (see what I did there? NaN rhymes with damn…sooo ..you get it right? ..right?).
It didn’t help that the lecture was hard to follow.

It’s unfortunate since I’ve been curious about DSP (Digital Signal Processing) and financial modeling for a long time, plus time series is one of the topics in this course I’ve had zero prior exposure to, so was ripe for some learning.

Hope isn’t lost though, since self-learners gonna self-learn. Planning on spending a few hours this break watching time series classes on YouTube to see what I missed.

Capstone Projects:

This week we started really thinking about our capstone projects.

During the last two weeks of the course we’ll be tasked with building an individual project that will serve as evidence for employers that we can actually do “data science”.

We were given access to a Github repo with an overview for building and presenting our projects, past examples, and how to come up with good project ideas.
Our initial ideas are due on January 5th, just two days after coming back from break, basically, that means we need to get crack-a-lacking choosing what we want to work on!

Earlier in the week we had an informal meetup with Galvanize’s Web Development students who, as luck would have it, are building and presenting their capstone projects at the same time as us Data Science students.

That means, the two groups have the rare opportunity to collaborate on projects. Data Science students doing the data and algorithmic parts on the back-end and Web Dev students creating the slick front-end.

I thought this was a great idea and that it would be really cool to work with a web dev on a project, but during the meetup it was clear pretty much everyone in both groups had no thoughts on what they wanted to do, and the discussions were very, very vague. No surprise.

It was probably too early for that kind of powwow.. a better time would have been one of the days shortly after both groups had to submit their initial project proposals.
Or better yet, create an open document where students write their name and idea, and if someone is interested in working with you on your idea, they can match up with you in the doc or get in touch some other way. That could be something..

For me, I’ve been chewing on a few ideas for 1 or 2 weeks now, and with some help, have narrowed them down to 2 that I think are pretty cool.

My main goals for the project are to be able to answer a real problem and create something that will actually be useful.. probably by turning it into a web app.

After brainstorming for a while, you’d be surprised how tough it is to find ideas that fit these constraints (mainly the “actually useful” part, lol)!
I ended up scrapping a bunch of ideas that seemed cool at first, because after taking a step back, they sounded more like toys than something with business value.

I’m hoping one of my two ideas passes the project proposal test, cuz I highly doubt I can come up with a third!

Holiday Break:

No Galvanize classes next week, so probably no post either.

It’s a 10-day break (..though I’m sure it’ll feel like it went by in all of two seconds), so I’ve made this VERY accurate breakdown of how I plan to spend it:

Alt

Happy Holidays ya’ll!