22 Sept 2021
Many students are interested in finding solutions for social problems. Following the success of this year’s Summer School on the topic, which was co-organized by MCML researchers at LMU, preparations are now underway for 2022.
“Unfortunately, I have very little time – I have to attend a meeting in an hour,” says Utku Can Ozturk, smiling at the camera. Utku is a student at LMU, but he was speaking from Turkey, although he was not on holiday. In fact, he was up to his eyes in work. He also had other meetings scheduled for that day, and then devoted the rest of the day to the task of debugging the computer code he has written. Terms like data, numbers, codes and algorithms might not sound very exciting to most people. But Utku, a budding data scientist, was one of 16 hand-picked students from all over the world who were taking part in the12-week Data Science for Social Good (DSSGx) UK 2021 Summer Programme. As a fellow, he was brought into contact with project partners, who share his aim to make the world a better place – by finding solutions to some current social challenges.
The Summer Programme was delivered by the University of Warwick (UK) in collaboration with researchers from LMU/MCML under the Data Science for Social Good UK chapter (DSSGx) of the DSSG Foundation.
What exactly is meant by the term ‘social good’ in this context? Professor Bernd Bischl, chair of statistical learning & data science and co-director of the Munich Center for Machine Learning (MCML), in his role as DSSGx 2021 LMU’s programme director explains. “All non-for-profit organisations and government bodies that are engaged in community-based schemes and wish to accelerate their implementation can submit projects for consideration.” Each carefully selected project is then worked on by a small team of participating students - supervised by specially hired technical mentors and project managers. This year’s projects were concerned with detecting corruption in Paraguay, mapping the off-line population in Brasil & Thailand, prioritizing environmental complaints in Chile and improving economic forecast during times of crisis in Germany. The latter was in collaboration with the German Federal Ministry for Economic Affairs and Energy (BMWi). In the light of the COVID-19 pandemic, this project aimed at robustifying BMWi’s forecasts by developing a bottom-up forecasting model for predicting economic development in Germany on a regional level.
“In our project, we are collaborating with the International Telecommunication Union (ITU) and UNICEF,” says Utku. The goal of this project is to provide more people with access to digital modes of communication. Utku points out that those who are unable to log on to the Internet are cut off from today‘s digital world of communication and news, which itself already makes a significant contribution to education. His team began by collecting statistics, and with the aid of data on communications satellites, information on social media and surveys, they identified those regions that still lack access to the internet. “Our idea is to focus on providing local schools with access to the internet, partly because they are often the social centers of the community or village,” Utku explains. This would also enable people in the surrounding districts to benefit from digital media. “We chose to concentrate on Brazil and Thailand as representative targets. On the basis of the data we had collected, we then wrote a program that makes it easier to collect and process the data, so that we are now theoretically in a position to assess the availability of Internet access in any country or region.”
„The case-based intensive training allows fellows to experience all the steps from project scoping to implementation, preparing them for the job market like few other programmes.”Prof. Dr. Bernd Bischl
One of the aims of the Summer Programme is to nurture a greater awareness of social issues and projects among highly qualified students and graduates. The teams spent three months writing code, pre-processing data and evaluating models, being supervised and guided by experienced technical mentors and professional project managers. They interacted with their peers on a daily basis and regularly discussed their progress with the project partners. “Through the Summer Programme, technical mentors and project managers provide students with broad and high-quality training including technical, project management, ethics and team-working skills. The case-based intensive training allows fellows to experience all the steps from project scoping to implementation, preparing them for the job market like few other programmes.”, says Bischl.
“The projects are selected primarily on the basis of three factors,” Bischl explains. “The social relevance of the project is tremendously important. – After all, the goal of the Summer Programme is to change things for the better. In addition, we focus on the international character of the whole venture, and we choose project partners who are in urgent need of our support. The same strategy applies to the choice of students. “Our projects are both international and diverse, and that is reflected in our student intake,” says Bischl. More than 470 students applied for the 16 places available. In addition to personal motivation, the major selection criteria were the technical qualifications of the applicants, and the wish to assemble the most diverse and international group possible. Because many students would normally have to earn money during the holidays, “participants in the Summer Programme receive a stipend, so that all suitably qualified applicants can apply, irrespective of their financial resources.”
The 12 weeks are now over. After a stimulating keynote by Ben Goldacre, the students got to present their exciting results at the closing ceremony, called Datafest. The written codes will be published soon, and will be available as open-source codes for use by developers all over the world. Hence, the efforts of the fellows will not remain restricted to a small group of users. Instead, all those interested will have the opportunity to work with their codes – or to improve them. Utku is looking forward to the subsequent phases of the project. Even though he will not be able to actively support it (the next semester is coming up!), he will of course continue to keep an eye on its progress. The next big step will be to bring the results to the attention of various governments, followed by applications for funding for its practical implementation.
Planning for the next Summer Programme has already begun. “We consider this year’s joint organization of DSSGx with the University of Warwick a huge success and are already making plans for next year” says Bernd Bischl. The Munich Center for Machine Learning is definitely interested in continuing its work on DSSGx in the future.