From smarter cities to epidemic control: algorithms can help
Where should you plant ten trees so that as many city-dwellers as possible can enjoy them? If a smart algorithm knows how people move through the city and where there are already trees, it can calculate the optimal solution. Data scientist Mitra Baratchi makes this possible. Her students are now using this method to study corona policy.
A lot of data is publicly available. Anyone can use the online satellite data portal or data.overheid.nl. Enormous quantities of data are also available through smartphones: wifi usage tells us where it is busiest at a festival, for instance. Although we are suspicious of this, it can also do a lot of good, helping us develop the best policy to tackle traffic congestion, for instance.
Mitra Baratchi looks at how to glean useful information from data. Her work covers various different fields and involves cumulating the right data for algorithms that she develops that help computer programs learn. These programs can then automatically generate solutions.
Where to open a shop
Within the Automated Design of Algorithms (ADA, part of LIACS) research group, Baratchi focuses on spatio-temporal data: data that is collected across space and time. She developed the Urban Computing course for master’s students to open their eyes to the possibilities. ‘Imagine an entrepreneur wants to open a shop in a town. He wants it to be in a place with lots of passers-by in the daytime. The city management, however, wants to spread things out rather than have all the shops lumped together in one place.’ For the international NetMob Future Cities Challenge, Baratchi’s students developed an algorithm that uses data about shop locations and mobility to propose the best locations, locations that will satisfy both entrepreneur and city management.
Baratchi wants to keep on improving the algorithms that she designs. She wants them to recognise patterns, while taking account of all the relationships between space and time. ‘An important geographical law is that everything is related to everything else, but near things are more related than distant ones. For instance, the temperature in a city is often closer to the temperature in a nearby city than in a city that is further away. I try to make my algorithms smart enough to understand and use that information.’
Optimal number of grazers in the Oostvaardersplassen nature reserve
Baratchi also wants her algorithms to be able to use data that comes from different sources, has been collected with different techniques and is of a differing quality. The algorithm has to be smart enough to adapt to the data that it is given. Baratchi is supervising a PhD candidate who is researching nature conservation in the Oostvaardersplassen nature reserve. ‘We have to generate our own data here because satellite data about the development of the vegetation is not enough. We have therefore hung up cameras to follow the behaviour of the large grazers in the area. We hope to inform decisions about the desired population growth and interventions.’
Are pupils thriving?
Although data-driven policy hasn’t yet taken off, as Baratchi would like it to, she is already able to solve real-life problems. For instance, about the efficacy of special education for vulnerable children. ‘Within Leiden-Delft-Erasmus, we have just started the Centre for BOLD Cities.’ [Big, Open and Linked Data, ed.] ‘We are investigating whether children with autism thrive in the special education concept or at a regular school. The children wear sensors on their clothes at breaktime to measure whether they are close to other children or more isolated.’
Where and for how long the children are in close proximity of one another gives an indication of their social interaction. This information can also provide ideas on how to improve the school playground, for instance, to create a more inclusive environment. A questionnaire alone would never provide such accurate and detailed information.
What is the best corona policy?
Students on the Computer Science programme with specialisations such as Data Science and Artificial Intelligence first learn from Baratchi which urban problems could be solved with data. Then they learn about which data they will require and to think creatively about where to obtain this data. The final step is learning to design algorithms that can extract the relevant information from all this data.
This academic year, Baratchi had decided that her students could look into corona, a good example of a spatio-temporal process. ‘They started thinking about how to capture the epidemic in a model in order to reduce the uncertainty about it. They were looking at different countries’ policy so that they could learn to predict from the results of different measures which effect each policy decision has had.’ The students started in February, so they couldn’t do all that much before the University closed, says Baratchi. However, it is clear that her research field is very promising indeed.
Text: Rianne Lindhout