13 Boats and a Castle, Airbnb in Boston
In December, in an effort to make the company more transparent, Airbnb released ~170,000 rows of anonymized listing data.
This is not that data.
Nonetheless, we’ll use VQL to look at public Airbnb data and find some interesting things. For example, the listings with the highest cleaning fees only have a cleanliness rating of 3 or 4 out of 10. And, the hosts with the most listings were renting rooms years before Airbnb was around. (Watch a video of the analysis here.)
To get Airbnb’s anonymized data, you need to schedule an appointment with their New York City office (but I couldn’t find any way to set up an appointment). So we’ll settle for some data recently scraped, by Inside Airbnb, a project which regularly scrapes Airbnb listings and reviews in 30 cities. Along with some neat visualizations using Multiple Coordinated Views, Inside Airbnb conveniently provides CSV files to download the data.
I downloaded some of the Boston data, inserted it into a PostgreSQL database and started analyzing it with VQL. The plug — VQL is a data analysis tool designed to make complex analysis of large data simple, without having to write code. Check it out at getvql.com.
The data was scraped on Oct 3, 2015 and contains 2,558 listings and 43,123 reviews.
The neighborhood with the most listings is Boston’s South End with 251 listings, followed by Jamaica Plain (of infamous J.P. Licks ice cream) with 240 listings.
Overall, the average price per night in Boston is $181. The most expensive neighborhoods are South Boston Waterfront, averaging $310 per night, and the six listings in the leather district, next to South Station, averaging a whopping $431 per night!
Looking at property types, 76% of listings are made up of apartments and 17% of houses. There are 3 “dorms” available, including an MIT Grad student’s dorm room and one pretty nice looking bungalow. Interestingly, there are 13 boats available and one “castle” in Roxbury, which has an average rating of 98!
The hosts with the most listings in Boston are short term rental companies, Seamless Transition with 70 listings and Maverick Empire with 43. Are these companies products or predecessors of Airbnb? The domain SeamlessTransistion.com was registered in March 2005, 3 years before Airbnb launched, and has had a website listing short term rentals since at least June 2007, one year before Airbnb launched in 2008. Maverick has had a website with short term rentals since April 2001 (Flash intro included!), seven years before Airbnb!
Since this data is publicly culled from Airbnb, we don’t know when listings are occupied, but we can try to use reviews as a proxy. For example, the most common days to write reviews are Mondays, likely right after weekend stays. Similarly, in May 2015, the most popular days to have posted reviews were May 19 and 30th, the days following Boston University’s and Harvard’s graduations. In April 2015, reviews spiked following the Marathon Monday long weekend.
Cleanliness vs Cleaning Fee
Ranging from $5.00 to $100, 63% of listings have a cleaning fee. 44% of listings with a cleaning fee have a cleanliness rating of 10 out of 10 vs only 37% of listings without cleaning fee. Ironically, however, the listings with the highest average cleaning fees have a cleanliness rating of just 3 and 4! There’s just one listing with a cleanliness rating of 3, and it comes with a $50 cleaning fee.
It would be exciting to get reservation data to confirm that spikes in posting times correlate with spikes in occupancy. Regardless, it’s interesting to see spikes in posts following weekends, holidays, and big events.
I was surprised to learn that the biggest listers on Airbnb, were renting short term rentals well before Airbnb even launched.
Lastly, it looks like cleaning fees don’t relate to cleanliness and are just a good way for hosts to charge a bit more!
VQL — Easy Data Analysis
The analysis above was done using VQL, a tool to ask questions of data without having to write code. If you’re interested in trying VQL, send us a line at email@example.com. To see how this analysis was done, check out a video or a really long screenshot.
Thanks to Alex and Simon for reading drafts of this post.