some more data processing..

I recently ran into some fascinating data from the online expense aggregator Mint.com that showed the average amounts people, (at least mint.com users), in RVA spent on food and dining. A bit of fiscal voyeurism perhaps, but it was kind of interesting to track how people spent money and fun way to waste time while wasting time.

How the data was processed: there is all kinds of data showing all sorts of expenses, ranging from the size of a check at fas mart or uppys to how much people spent at costco on average ($133.89). I sifted through the 126 places initially listed and...

1. removed all fastfood places
2. removed all chains (avg Chipotle check - $10.74)
3. removed all gas stations, grocery stores and other stores (eg. wine stores. ABC - $31.42)

what remained ~ 40 RVA specific 'eating places' ranked in order of an average check at each (one outlier omitted, see below). I wish there were more "restaurants" in this data - insufficient or no data is expected and probably the (sparse?) mint user demographic skews a certain way. Click to embiggen ...fun w/ excel - two versions.(sorry I'm not a graphic artist !)


It is also fun to read into the data and try and figure out some of the costs. However, it is also easy to be misled by most numbers - there is no normalization with respect to number of people on a check, whether it is for lunch/dinner/just drinks, number of times this happened (statistically significant n?). For example, the average check at Starbucks is listed at $6.16. Are people getting really expensive soy, venti, decaf blah blah or just paying for two (or more) people etc etc...
The outlier that I skipped was Bank with an "average check" of $347.18...I would guess the n on this is really low, skewing the number. Similarly, it is easy to understand why places like Ipanema (or others in the VCU area have lower values, where a split check might be the norm). Understandably, data from larger cities was a bit more comprehensive and with nicer distribution across restaurant ranges.

This data is faceless but its only a matter of time such data is mined extensively, especially with sites like blippy with social networking of expenses becoming popular.

Which brings me to wonder - is it useful for a customer to know what the average size of a check at restaurant X is (this is quite distinct from knowing what the price of an average entree is at restaurant X)? does it help the business that people know?