Serious Data Science
Most of this is proprietary! But here’s whatever I can talk about more publicly:
At Stitch Fix, we have lots of images of merch. Knowing what color they are is important. Here’s a two part blog post (one, two) about how we framed the problem and solved it.
Very good internet nonsense
Buying stuff is great! But what if you don’t want any more stuff? Enter Pokémon NØ: bring your own barcode scanner and collect pokémon by scanning anything in the store. Created at NØ School Nevers, 2021.
These memes are very funny. I don’t know why. So I made a Twitter bot that makes a lot more of them. 2014. code
Radiohead or Horse_Ebooks?
Alas, @horse_ebooks is no longer a seller of equine .epubs, but for a while it was one of the most wonderfully cryptic things on the internet. This is my little tribute to both the old horse and an equally cryptic band. 2012. Test your own knowledge.
Neighborhoods and cities
Generating neighborhood guides from social media data
Imagine you’re traveling to a new city and trying to rent a place to stay. What neighborhood will you stay in? What area feels right for you? I built neighborhood guides based on the social media that people are posting there already. PhD Thesis, proposal, the much more interesting talk, and the ICWSM 2016 Citylab workshop paper.
Why do people geotag?
People post their locations on their tweets, checkins, photos - why? We went to investigate that, and ended up finding some other weird nuggets in the world of geotagging today. ICWSM 2017 - pdf
Our House, in the Middle of Our Tweets
If you have someone’s geotagged tweet stream, how well can you find their home? About 80% of people within 1km. ICWSM 2016 - pdf