DSSG Summer Fellowship 2023 in Warwick
The 5th edition of Data Science for Social Good delivers practicals tools for nonprofits and government organisations
| August 10, 2023
This article was originally published on the September 2023 issue of the IFORS Newsletter.
Data Science for Social (DSSG) is an initiative to push data science as a driver for positive social impact. Every year, the DSSG summer fellowship trains and supports the next generation of data scientists across different programs in the US, the UK and Germany. Fellows build software tools to support nonprofits in making better use of their data, often tasked with overcoming some of the biases in machine learning and artificial intelligence.
Since 2019, the University of Warwick has hosted its own DSSGx summer fellowship. This year’s edition brought together 16 fellows from 9 different countries between June and August. Their academic backgrounds vary greatly to ensure that the challenges they face are tackled from different perspectives.
Over the twelve weeks of the program, the fellows work closely with project managers and technical mentors, who support them in making industry-tier software products for their partners. In addition, weekly seminars with scholars and practitioners encourage discussion on topics such as fairness in medical machine learning, and artificial intelligence approaches to detecting misinformation.
For the UN-REDD, a group of fellows was tasked with forecasting deforestation in the Amazonian region of Brazil. “Modelling deforestation is a big challenge because the data is very imbalanced. We are not only solving a classification problem but trying to predict an outcome,” says Dmytro (20), a software engineering student from Ukraine working on this project. “60% of the Amazonian Forest is in Brazil; if things go bad there, they go bad everywhere. Hopefully, we can help win the battle against deforestation.” This team has developed a data visualization tool to help policymakers fight the loss of the Amazonian Forest more effectively.
The EY Foundation, which supports young people from low-income backgrounds, is working with a group of fellows to identify students at risk of becoming NEET (not in education, employment, or training). The team has developed a machine learning algorithm that flags students at risk early on so that the five local authorities involved in the project can provide additional support for them. Nazeefa, (22) an electrical engineer from Pakistan, describes the impact they hope to have: “If you can predict [NEET outcomes] as early as possible, you can intervene and mitigate some of the mental health issues associated with it. That is the most important outcome of this project.” The team has developed a dashboard that can be deployed by the councils and used as a tool to identify young people at risk of becoming NEET.
Another team of fellows partnered with the Algorithmic Transparency Institute with the challenge of classifying social media posts from carbon-intensive companies using language and image recognition models. Allassan (26), a machine intelligence researcher from Cameroon, shares that “the goal was to automatically label social media posts by how strongly they emphasise environmental activities.” Their tool is designed to support the detection of greenwashing by firms in the aviation, automation, and fossil-fuel sectors. United Learning is a multi-academy trust overseeing around 100 schools in the UK. Their goal is to help their students make full use of their academic potential.
Another group of fellows was prompted with detecting pupils at risk of not fulfilling their higher education potential. “We are hoping to help schools in supporting students, but also making sure that their academic staff can make data-informed decisions”, says Kirtana (21), a social data scientist from India and the US. “Our biggest challenge was figuring out how we can maximize our use of the data United Learning provided, to produce the best possible result we can get, but without it being biased, which is a difficult balance to achieve.”
In addition to working on pressing social issues, the fellows get to develop new skills and get to know potential career paths; “I haven’t done much data science, as my background is mainly in software engineering. Doing this project was very fun and reassured me to pursue a career in machine learning,” says Dmytro. Nazeefa believes that this experience “is all about the people. All the diverse backgrounds, the different experiences... it’s the best part about DSSG.”
Allassan explains that “before, I wasn’t sure I was following best software developing practices. Through this project, and my technical mentors, I could learn about and apply best practices to my code.” For Kirtana, the collaborative aspect of this experience is the biggest takeaway: “DSSG has fully built from scratch my ability to work with Git to collaborate with others and make sure that the versions of what we’re coding are consistent”.
Students who have a passion for data science, care about the social good and love to work in an international team of top talent, are encouraged to look out for next year’s application deadline at our website.
Charities or government organisations who would like to benefit from and work with DSSGxUK may contact the programme by writing to dssg@wbs.ac.uk.