Using Data Science to Solve Social Science Problems

 How can data science be harnessed by social sector organisations to address key problems and unriddle overarching social questions?DataKind UK, an organisation that looks at how data can be used for social change and impact, recently held it’s Autumn DataDive, sponsored by SAGE Publishing, in order to help two not-for-profit organisations (WeFarm and OneWorld) solve their data problems.But why are events like this so important, not only for the organizations themselves but also for us, as the public whom these data analysis could impact? Furthermore, in this era of big data, what challenges do data scientists face when attempting to tackle these issues? We spoke to Emma Prest, general manager at DataKind UK, to find out all about what happened at the DataDive and the outcomes for the non-profits:Why is the work of DataKind so important to research and our analysis of it? I imagine our work is of interest to traditional research communities because our approach is so different, and yet sometimes answers similar questions. Our DataDives offer short, sharp injections of data science support to solve social sector organisations’ problems. Sometimes those problems can be overarching questions that affect all of us in society and could be treated as a long term research project, but we tackle it in a weekend, e.g. what is the level of youth homelessness in the UK? Other times the problems we address can be more operational, such as, can we predict which of our beneficiaries are likely to become repeat users at our foodbank?Our DataDive format offers a quick and dirty way of seeing what kinds of data analysis can help to tackle an organisation’s questions, and get them answers very quickly. It is also an eye-opening experience for the staff at a non-profit to sit next to and work with a team of data scientists over the course of a couple of days. Their data literacy often skyrockets throughout the process.I don’t think what we’re doing replaces traditional research. It is a new offering, one that hopefully excites and inspires people to do more with their data, and involves more beer and pizza!What are the key challenges that you think data scientists face?I think one of the hardest challenges for data scientists is communicating their work in a way that non-technical people understand. Explaining why they chose to address a certain question, how they did it and what the impact of that is for the organisation is essential when trying to engage the non-profit, but not something that comes naturally to all data scientists. The results you produce are only as good as your ability to communicate them.After the last DataDive, can you tell us more about the challenges that came up?There were the usual challenges when running a hackathon-style event: trying to rally a room full of strangers to deliver productive outputs; making sure volunteers feel they are contributing useful work; ordering the right number of burritos.We do a lot of preparation for DataDives to avoid some of these problems. Each social sector organisation is teamed up with three volunteer Data Ambassadors who help them to prep their data and hone the questions they are asking. The Data Ambassadors lead the project during the event and are responsible for coordinating the team of 20-30 data science participants who show up on the day. That said, the event is always creative chaos and that’s kind of the point.What are the next steps in how to address the outcomes of the event?We are in the process of holding wrap up meetings with the two project teams to sift through all the outputs from the weekend. We are thinking through which ideas and analytical approaches are most relevant to their work going forward.Sometimes organisations continue to work with a volunteer or two to implement work produced at the DataDive. Other times organisations come back to us and we run longer term DataCorps projects to deliver more ambitious data science solutions.