User-mining: Start at the gym, end up at a bar?

http://simplyryan.com/2010/07/27/user-mining/

by rwaliany • 16 years ago

96 22 16 years ago

The restaurant data is the only one that is at odds with intuition, and strongly so. Why would Thursday be 3.5x more popular than Wednesday? Without an explanation, I'm more willing to believe that there was a data anomaly.

rwaliany3• 16 years ago

The data suggests that Thursday night Tapas bar/restaurants are very popular and that is included under "restaurants" according to yelp.com.

drc19122• 16 years ago

Thursday is a _HUGE_ night at bars (and restaurants with bars) for college students. A lot of kids will take advantage of happy hour pricing on Thursday night and lax attendance policy on Friday morning.

This is anecdotal and I don't have data to support my claim (although someone can probably find something on Google).

kajecounterhack• 16 years ago

Yeah, thirsty thursday (college student here)

ahoyhere• 16 years ago

You're making a critical logical mistake - assuming a data anomaly when you don't know anything about the field, but you take a 5-second check with your own intuition, and can't find an explanation.

You've never heard "Thursday's the new Friday"? College kids and young employeds party hearty on Thursdays now. It started in the early 2000's.

Uncle Google shows that even the NYT knows about it: http://www.google.com/search?q=thursdays%20the%20new%20frida...

Remember: just because you can't figure out the reason intuitively, and immediately, doesn't mean there isn't one.

gaius• 16 years ago

This started in the City as far back as the 80s - everyone wanted to leave work early on Friday and go away for the weekend with their partners/families/whatever. So Thursday was the night for getting drunk with your cow-orkers. Casual Fridays comes from this time too.

philjr• 16 years ago

Genuinely not sure about that, but considering how this analysis was done, it's equally likely, that Thursday is the day that people would start talking about a restaurant ... i.e. looking for recommendations for the weekend or talking about going to a particular restaurant.

mahmud• 16 years ago

In almost every major venue I have frequented, Thursday night is considered "ladies" night. It's when they waive cover charges for women, luring them in, and in the process lure in more men.

revanthv• 16 years ago

This awesome.Having this kind of analysis really helps

drc19122• 16 years ago

Can you please tell us a little bit about the data/population that served as the basis of this analysis (ex: how big is total population of data set, how big is the subset of users who went to both gym and park, etc). It would help us better provide feedback and understand the your results.

rwaliany3• 16 years ago

The data is from March 15th to May 1st (~45 days). It is a noisy estimation based on foursquare and twitter location check-ins (estimated samples from a population of about 100k users in San Francisco, CA with a technology bias). I haven't analyzed user-specific data, such as who went to both venues. I am open to suggestions on future analysis.

spectre• 16 years ago

Any idea why there are two peaks for bar attendance at 8am and 10am?

charliepark• 16 years ago

It uses Twitter posts (sentiment) as well as actual Foursquare check-ins (location). I'm guessing people are waking up and either commenting on how they shouldn't have partied so hard the night before, or are saying something to the effect of "why'd I leave the bar?" / "I should move in to the bar" / etc.

xiaoma• 16 years ago

I'm amazed so few people are at the parks in the mornings. I always thought that they were crammed with people walking their animals, exercising, etc... from around 5:30 to 8:00 am.

I guess people there like to sleep in.

winthrowe• 16 years ago

I think this shows the skewed distribution of geolocation users, as compared to the population. I can't find the link right now, but there was a story posted recently about a survey finding less then 10% of the population had ever heard of geolocation services such as foursquare.

Herald_MJ• 16 years ago

Am i infer that half-way between wednesday and thursday (midnight) that the check-in rate for restaurants is half way between the two (~450)?

Silly choice of graph type.

rwaliany3• 16 years ago

Updated graph on gym usage by hour: http://simplyryan.files.wordpress.com/2010/07/gymhour.png

digitallogic• 16 years ago

Which is why I make it a point to go to the gym in the morning or at lunch if I sleep in. Absurdly over packed after everyone gets off work.

texasice• 16 years ago

Nice work.

festivusr• 16 years ago

Has any study been done about the likelihood of people to check-in with one of these social apps? If someone checks-in at the gym on Monday, are they going to be less likely to check in again on Tuesday? People I know seem to check-in when they are subtly bragging about what they're doing, so is it likely they'll check-in at a non-trendy restaurant on a Tuesday evening?

User-mining: Start at the gym, end up at a bar?

Related Stories