Four of our superb developers won the AutoTrader competition at HackManchester, a 24 hour coding challenge, this weekend. HackManchester is an annual event as part of Manchester Science Festival and is a type of Hackathon. A Hackathon is an event which a group of computer programmers and others involved in the software and hardware development process like front-end designers, work together in an intense environment on projects, in competition with other teams.
The team, GoodLuckToYou, consisted of our CTO Ben Gibbs; Mark Brewster; Matt Whiteley and Bruce Wilson .
The challenge we entered was the AutoTrader challenge, which was to make the most interesting use of the Greater Manchester open police data-set. So, with a tongue-in-cheek approach, our team created GAWI: the free, anonymous, and beautiful way to optimise your crime sprees!
The Techy Stuff
We used the GMP crime data for the last 12 months (over 360,000 reported crimes) and ran through them all to determine which crimes were resolved (about 36,000) based on their outcomes (a lot of crimes had no outcome data).
We then needed to collect these crimes into areas which made sense to a local person. The police crime data collects them into "Neighbourhoods" but these covered huge areas. However, each crime is also listed under the Lower Super Output Area. The problem was that the LSOA's are too fine-grained and named "Bolton 00AD" - which would make no sense to a local person.
So, we used Postcodes.io to extract the "Admin Ward", or electoral ward, which has names like "Rusholme" or "Crumpsall" - much better - and we joined all the crimes into these Admin Wards.
We then knew the number of reported and resolved crimes per area and could work out the GAWI score, or likelihood of being convicted for particular crimes in those areas. We were then able to do some geolocation queries to find the nearest areas and what their GAWI score was to suggest better places to commit the crime, or places you should avoid because you were more likely to get caught.
We also used the OpenWeatherMap API to provide the 5 day weather forecast for the chosen area, Twitter to embed the local Police Neighbourhood feed, and Wikipedia's API to get the blurb about the Neighbourhood to provide context for the criminal!
We used the Police's API to extract the Police officers in that neighbourhood, and the event they had planned.
Then we threw all this in the beautifully designed inteface Bruce provided and built and GAWI was born! We used Heroku's auto-deploy feature with the Github repository where we stored our code.
We had a lot of fun getting into the mindset of a criminal and wondering what data they would find useful. We worked really well as a team, each taking over as the others "crashed" from tiredness or sugar rushes! The great part of working together at Volcanic is that we push each other to get the best out of each other and by doing this we've achieved so much.
We won a Raspberry Pi 2, which is a tiny computer for projects, and £30 Amazon vouchers each. The mystery prize was a Lego Creators Ferrari. Needless to say we're very chuffed :)
You can see the award segment we won here.