Hackathons are a way to bring together groups to work intensively on a software project over a limited period of time, usually a day to a week. Often competitions, teams present their work at the end of the hackathon.
Hackathons are all about collaboration, focus and hard work. The time period of any hackathon is limited, so they tend to be very focused, and very intense. People with different areas of expertise are brought together, and the object is to collaborate to solve problems. Hackathons often have a particular focus, such as a problem that someone wants to solve, or a piece of software or app that needs new input.
Hackathons are not (necessarily) about the product. This may sound surprising, but many of the big hackathons are not really about the products. Hackers have a term for the software created at hackathons: vaporware, because it evaporates after the event. In-company hackathons have often come up with more sustainable ideas, but many public hackathons tend to be part social events, part learning experience, and part recruitment and networking opportunities.
One of the best things about hackathons may be the networking opportunities. Hackathons attract large numbers of techie people. This means that they are a great way to meet other people who are interested in the same kind of thing, and also who may be able to provide expertise and/or work further down the line. For companies, they are a good way to test possible recruits in advance of hiring.
Data science hackathons focus on using data science to solve difficult problems. Data science hackathons are slightly different. Rather than bringing together programmers to create new apps, they bring together data science experts to try to understand, and work out how to address, big, challenging problems. The organisers identify the problem, and provide the data. Challenges might include, for example, understanding why there is such a big educational achievement gap between students from different ethnic backgrounds, or solving societal problems in Sweden.
Data science hackathons are a team effort—and need the right team. One of the most important ‘inputs’ to a data science hackathon is the right teams. The members need a range of skills to enable them to work together and complement each other. In ordinary hackathons, competitors pitch ideas and form teams as part of the process; in data science hackathons, the teams often enter together, although this is not always the case.
Data science hackathons need input from experts who understand the data. As in business, data science hackathons need data science experts. But these experts often need help to understand the data that they are analysing. If you don’t understand your inputs, you may not ask the right questions or come up with useful hypotheses, and you certainly won’t understand the answers that emerge from the process.
Clean data is an essential input into the process. Prior preparation and planning is vital for data science hackathons. The data supplied needs to be pulled together into a usable format. It also needs to be as clean as possible: high quality data is essential. If the hackathon organisers don’t do this cleaning ahead of time, an awful lot of hackathon time is going to be spent on it to ensure the results are worth having.
Geographical information can provide incredibly useful insights. Information from Geographical Information Systems (GIS) can often be a helpful input into data science hackathons. It may add a new dimension to the analysis, and also show insights that would not otherwise be available about the connections between variables. If possible, it is worth making this data available.
In data science hackathons, useful results may emerge from all groups. In ‘ordinary’ hackathons, groups are likely to be working on very different ideas. In a data science hackathon, the groups are more likely to be working on the same issue. Although there may still be a winning team, the organisers will probably obtain useful ideas from all the groups.
It may help participants to look at some resources in advance. Prior preparation and planning is not just important for the organisers. Participants can also benefit from doing a bit of preparation work, such as learning to use a basic data science language like Python, or running through a list of useful resources and trying them out.