I won best use of data at NASA’s SpaceApps 2017 in NYC with a project to unify NASA’s ~38,000 open data file formats and APIs.
2017 American Association of Geographer’s Annual Meeting
I’ll be speaking at the 2017 American Association of Geographer’s Annual Meeting in Boston about Dstillery’s research into modeling political support during the 2016 Iowa Presidential Caucuses.
Geography at Internet Scale
Geography at Internet Scale is a talk I gave at Dataengconf 2016 on the top things data engineers and data scientists needs to know about my favorite subject:
Robot Donald Trump Twitterbot
I created the Robot Donald Trump twitter bot in a few days during summer 2016. It uses Markov chains, trained on Trump campaign speeches, random songs, and crazy conspiracy theory sites to generate a tweet every 15 minutes. Automated grammar-correction functions turn the gibberish Markov chain output into something resembling English. Noun phrase chunking finds subjects to turn into hashtags and Sentiment analysis is used to add Trump’s characteristic terminating mood statements. It also occasionally adds a link to a site related to the tweet.
As of Dec. 2016 it was getting 100,000+ hits a month.
I shut it down in Jan. 2017 because it just got so depressing – passion projects should be fun!
Here’s it’s latest tweets:
Continue reading
2016 Iowa Caucus Research
I really love political data – so getting to use Dstillery’s billions of events and heavy duty machine learning pipeline to make a marketing piece for the start of election year 2016 was right up my alley. Together with my colleague Pete Ibarra (who I also collaborated with on our …
Using Digital Signals To Measure Audience Brand Engagement At Major Sports Events: The 2015 MLB Season
Peter Ibarra, my colleague at Dstillery, are both big baseball fans. One day we ideally wondered: ‘Who actually goes to baseball games?’ One of the great things about having so much data and device graph is that we can answer that question. This paper was best Business of Sports paper …
Mentioned in the New York Times
My work on Sapling Detector at the 2015 New York Museum of Natural History ‘Hack the Dinos’ hackathon got me a quoted in a New York Times article:‘Hack the Dinos’ Helps Paleontologists
Hack the Dinos
Carleen Pan, Karol Zięba, Carol Lin and I won best Visualization at American Natural History Museum’s Hack the Dinos Hackathon with our project, Sapling Detector, a way for paleontologists to automatically convert images of phylogenetic trees into human readable text formats using computer vision and deep learning techniques.
Interview with Geomarketing.com
Geomarketing.com interviewed me on the occasion of Dstillery’s launch of our new spatial filtering system – which made our geodata among the cleanest in the adtech ecosystem:
Dstillery Adds Self-Serve Ad Tools, Explores Ways Of Weeding Out Bad Geo-Data