dataaa.net

Last weekend I was fortunate enough to be one of 40 or so people to attend Belfast's "Urban Data Hackathon" weekend. Organised as part of Belfast's Big Data Week by Kainos Software, and supported with data from Belfast City Council, TomTom, TourismNI and various others, it challenged people to dig into some open and proprietary datasets to learn more about Belfast and to "Make Belfast Better".

We split into teams, and formulated questions to answer with the data we had been provided: a variety of information about recycling, events and vacant storefronts from Belfast City Council, Twitter stream analytics from TourismNI, route data from TomTom and others.

24 hours later, we presented our findings to a panel of judges, who picked a winner: a team which took data on household recycling in Belfast and came up with an idea to turn recycling into a competition, pitting one area of Belfast against another in a recycling war, for the glory of winning a gold badge for their area. Their idea of trying to motivate people to recycle was appreciated by the judges, although they did not use the data in any significant way. The second placed team also looked at the recycling data, but really dug into what was available, and pulled in other data resources (census information, weather) to try to understand the factors involved in recycling yield. There was not enough time to really produce a decent model, but their initial attempt found some minor correlations with different factors such as religion, gender, and whether or not you were German (Germans in Belfast tend not to produce much recyclable material, apparently...!). The third placed team looked at the position of vacant lots in Belfast city centre, and suggested using the vacant lots for start-up companies. They identified 44 'prime' lots that could be used for this purpose, but apart from that, did not really dig into the data or offer a viable solution in terms of cost effectiveness.

And so the weekend was a mixed bag, with some teams doing some excellent data analysis and not being rewarded for it, and other teams being very cursory in their data analysis but having good business ideas. It opened my eyes a little to the business side of data which I had not fully considered before.

My chosen project for the weekend was to analyse Twitter stream analytics for the @DiscoverNI Twitter account. Using the permalinks for all the tweets from the past year, I identified what I called @DiscoverNI's "Twitter Champions"... those followers of @DiscoverNI with a large following themselves who retweeted @DiscoverNI's content to their own audience. By identifying the most influential of these people, I suggested that they could be targeted and encouraged to continue sharing @DiscoverNI content through invitations, discounted tickets, etc. to @DiscoverNI events throughout the year. I did all the scripting in R, and was continually frustrated by the frequency and volume limitations on the Twitter API...

Overall, a good weekend with some interesting people and some fun challenges. The staff at Kainos did a great job of organising and facilitating the 'hackers'. Hopefully there will be more of these events to come in the future!