Rediscovering quantitative analysis.

July 28, 2008 10:58 AM

How can I get back in touch with my quantitative skills?

In high school I was a big math/statistics person, but in college I switched to social sciences and never looked back, and (despite what they might claim) the social sciences courses never really required you to do anything with actual numbers. So, cut to many years later, it's been years and years since I've had to do any serious data analysis, and...I kind of miss it.

My question is, how can I resharpen my ability to work with numbers, especially datasets, when it's just for fun? It's a skill I don't want to lose completely. Any good online courses or books? Other ways I can commit myself to working with data in my non-work life? Other recommendations? I really have no idea how to tackle this admittedly vague goal, but when I think about how rusty my ability to think about and analyze data has become, it bums me out.

Thanks for your help!

posted by EnormousTalkingOnion to education (7 comments total)

Guesstimation looks to me like a fun book when I've glanced at it in the bookstore. Maybe not exctly what you're looking for, but could sharpen your ability to deal easily with numbers and data.
posted by DarkForest at 11:18 AM on July 28, 2008

When you haven’t done data analysis (or any other technical thing) for a while, it’s amazing how much you forget.

I’d start off with the Cartoon Guide to Statistics for an overview of basic stats. Then I’d pick up a copy of Econometrics: Using Monte Carlo Simulation with Microsoft Excel. This book starts with correlation and works its way up to multiple regression, and beyond. The nice thing is that is uses only Excel (so you won’t have to go out and buy SPSS, etc.). Also monte carlo simulation is a good way to get an understanding of how statistical estimation works. I’d read Peter Kennedy’s book A Guide to Econometrics at the same time. It also promotes the monte carlo pedagogical technique of learning statistics.
posted by Jasper Friendly Bear at 11:30 AM on July 28, 2008

Also, as a refresher, there are two courses from the Teaching Company you might want to view at the same time you’re relearning your stats (for more of an immersive experience): What Are the Chances? Probability Made Clear and Meaning from Data: Statistics Made Clear. The Teaching company quite often has steep discounts on its products (so keep watching the website).
posted by Jasper Friendly Bear at 11:42 AM on July 28, 2008

Well, you've hit on the main problem, which is that you have to use these skills to keep them sharp, and that's hard when it's not your day job. You need to find some kind of project to work on that's interesting enough to keep your attention. Maybe the MeFi Infodump would give you some ideas. Or maybe you're a baseball fan, there's tons of data out there if you look around. E.g. A lot of academic data can be found with some digging as well.

Seconding the endorsement of the Kennedy book, it's the most accessible stats textbook I've encountered.

You can use R if you want more powerful analysis tools without spending money, but it's, er, not the most user-friendly interface ever developed. Steep learning curve. Plus it kind of sucks for data management, which is 90% of the battle.
posted by shadow vector at 11:53 AM on July 28, 2008

I think the people saying you need to use the skills regularly are spot-on; so the question becomes, how to do that without getting bored?

Depending on exactly how rusty your skills are, maybe you could volunteer someplace that could use that kind of work done? Or just come up with goofy projects; I once decided to do a survey and statistical analysis of all the cars in a local shopping center's parking lot, for instance, and posted my results online so that all my friends could point and laugh at my dorkiness. Hm, I guess these days that would be a blog thing -- maybe you could blog about goofy statistics projects you devise?

Whatever you choose, creating a concrete goal/project will of necessity guide you to the appropriate tools.
posted by obliquicity at 12:27 PM on July 28, 2008

You know, I've been facing a similar quandary myself.

I recently dusted off a book from grad school on how to value companies. (It's an M&A-focused book, and very technical, but I like that.) I'm going through it chapter-by-chapter to exercise my atrophying quant skills. A textbook is good because it has a clear set of learning goals, end-of-chapter exercises, summaries, and other pedagogical tools.

For me, there's some slim possibility I'd use this again in a professional capacity, but I like it because valuations of shares in public companies can also be done for fun.

I say "fun" rather than "big profits" because in general I think it's hard for an individual to be smarter about this than big institutional investors.

Still, sometimes certain shares seem out of whack, and I'd like to build a model of "what would you have to believe" for a current stock price to be true. Any share purchases are strictly gambling/entertainment, but this also acts as a nice incentive for me to re-learn some quant skills.
posted by CruiseSavvy at 4:28 PM on July 28, 2008

I will definitely check out the Kennedy book. I love obliquicity's suggestion too - some sort of site devoted to pointless statistical analysis of my everyday life...

Oh, and DarkForest, FWIW, I actually picked up Guesstimation a few months ago after it caught my eye in the bookstore, but I was EXTREMELY disappointed. It was like a 3rd grade word problem textbook. Not recommended.
posted by EnormousTalkingOnion at 7:10 AM on July 29, 2008