Comments Off
Renowned physicist Enrico Fermi’s name is synonymous with a type of estimation problem often illustrated by the classic question, “How many piano tuners are there in Chicago?” Finding a “Fermi estimate” of this number would typically involve multiplying a series of rough estimates, such as the population of Chicago, an approximate number of households owning pianos, the frequency with which a typical piano might be tuned, and so on. It’s unlikely that anyone would arrive at a precise, correct answer through this method, but a Fermi estimate should at least be able to generate an answer that is approximately the right order of magnitude.
A Fermi estimate usually seeks to measure a quantity that would be extremely difficult, if not impossible, to actually measure. “Piano tuners in Chicago” may have fallen into that category several decades ago, but as Wolfram|Alpha can now demonstrate, things have changed:
We recently overhauled our data on jobs and salaries in the United States, adding Bureau of Labor Statistics (BLS) data on more than 800 detailed occupations at the national, state, and metropolitan area levels. Which means Wolfram|Alpha can’t quite get you to an exact measurement of the number of piano tuners in Chicago (and presumably, many of them must at least dabble in other instruments), but it can come surprisingly close.
Wolfram|Alpha can also compute a number of interesting statistics that aren’t obvious from the source data, such as the fact that Chicago has quite a high density of musical instrument tuners and repairers—roughly 2.3 times the national average workforce fraction for this occupation—and that their median wage is roughly 1.3 times the national average. And it can also provide helpful context for any occupation, computing employment and wage information for related jobs and sub-specialties, according to BLS classifications.
You can also perform all kinds of interesting comparisons, of course: try asking Wolfram|Alpha to “compare producers and actors employment in California”, for example, or “garbage collectors vs waiters salaries in New York City”. Or if you’re contemplating a cross-country move, you might be interested to see a comparison between “computer programmers salaries in Seattle vs Philadelphia”.
And if you need to access salary and job-related data often, you can create your own Wolfram|Alpha Widgets tailored for specific jobs and regions. You can easily customize widgets, like the one below, and embed them in your website and share with your social networks.
Is that yearly change a known bug? Looks like it should be -20%, not -0.2%.
You’re right, Jamie. What a stupid mistake for us to make. A fix is in the works. Thanks for pointing this out.
or perhaps -60 a year was the mistake ? -20% seems really strange
That’s fascinating stuff!
Looking at the data on garbage collectors, Wolfram says they’re working 587.1 hours per week. No wonder they get paid more than waiters. 🙂
Philip,
Thanks for pointing this out. You’ve uncovered a nasty bug in our 2009 data. We’re working on a fix now.
I keep hearing all these great and wonderful things about WA, but coverage of non US data seems to be pitiful. Runnng any of the queries from your example with London instead of [US City] results in “WA still doesn’t know what to do”
#fail
Hi Fritz,
When we decide to add a “backbone” of sub-national data in a given topical area, U.S. statistical offices do tend to give us the most bang for the buck — in terms of providing the best combination of detailed current data and deep historical time series for a very large population. But incorporating non-U.S. information is definitely a priority for us, and as the UK opens up more and more national statistical data, we’ll do our best to teach Wolfram|Alpha how to compute it.
In 2009, someone had seen 44 RPTs in the Chicago chapter and 14 in the North Shore chapter, for a total of 58.
Counting RPTs probably gives you a lower bound, since to be an RPT one must pass a series of exams and pay non-trivial annual dues. But the actual number of piano tuners is most likely higher, for a couple of reasons:
* Some RPTs may choose to not have their information listed in the PTG directory, though since it amounts to free advertising, I would guess this to be rare.
* Some people who may be considered piano tuners (using whatever definition you want) may not be RPTs. This may be due to the cost and difficulty of the exams, the annual dues, or any other reason. I am not sure what this number is, but I would be surprised if the number of non-RPT tuners was more than 3 times the number of RPT tuners.