Mateusz Śmigielski, a graduate of our Faculty, was honoured in the 23rd edition of the Wrocław Magnolia competition. In this year's edition, graduates of Wrocław University of Science and Technology received a total of five awards, including three first places. Congratulations!

The competition is addressed to graduates of public and private universities in Wrocław. Their master's theses were expected to focus on improving the broadly understood quality of life of Wrocław residents and to address such topics as environmental protection and human health or the development of green areas. It is organised by the Wrocław Academic Centre.

The eighth place went to Mateusz Śmigielski, a graduate of Trusted Artificial Intelligence, for his thesis “Application of Artificial Intelligence in the Analysis and Prevention of Failures in the Municipal Water Supply Network of Wrocław”. It was prepared under the supervision of Dr. Tomasz Szandała from our Faculty. The winning graduates are from the Faculty of Environmental Engineering: Michał Dębczyński in the first place, followed by Weronika Pitra and Weronika Tosyali.

In his thesis, Mateusz Śmigielski combined the topics of urban infrastructure, open data, and machine learning. He emphasises that from the outset he was careful to make his research practical. The inspiration came from observing how urban infrastructure operates. Living near the MPWiK water utility in Wrocław, he witnessed the scale of activities related to the maintenance of the water supply network.

The author used publicly available data about water supply failures in Wrocław. However, the data was not ready for analysis. It was in the form of short text messages, containing, among others, the date, time, type of event, its expected duration, and addresses affected by the problem. Therefore, the first stage of the work involved cleaning the data and automatically extracting addresses from it.

– The goal was to automatically extract the street name, addresses, and house numbers from the text. Each message passing through the machine learning model returned specific data, and the address-extraction accuracy exceeded 90 percent – explains our graduate.

As a result, it was possible to automatically process over 80,000 messages covering a period of two years, assign them geographical coordinates, and create a map of failures. According to the author, simply plotting the events on a map was not enough, because failures occurred almost everywhere in the city.

– I wanted to check if there are areas more prone to failures, for example due to older infrastructure. That is why I applied clustering methods. They identified a dozen or so areas in Wrocław where failures occurred particularly often – says Mateusz Śmigielski.

The final stage involved an attempt to predict future failures using machine learning models. As the author points out, this element still requires further development, but the analysis alone has already identified locations that require additional attention. Such a tool can be passed on and used to indicate areas where the number of failures is above average – he adds.

Full list of winners:

• 1st place – Michał Dębczyński (Wrocław University of Science and Technology, Faculty of Environmental Engineering)
• 2nd place – Weronika Pitra (Wrocław University of Science and Technology, Faculty of Environmental Engineering)
• 3rd place – Weronika Tosyali (Wrocław University of Science and Technology, Faculty of Environmental Engineering)
• 4th place – Helena Nowakowska (Academy of Fine Arts in Wrocław)
• 5th place – Laura Jonderko (Wrocław Medical University)
• 6th place – Hanna Witoszek (University of Wrocław)
• 7th place – Jakub Słowikowski (Wrocław University of Science and Technology, Faculty of Environmental Engineering)
• 8th place – Mateusz Śmigielski (Wrocław University of Science and Technology, Faculty of Information and Communication Technology)
• 9th place – Maria Grzegorzewicz (Academy of Fine Arts in Wrocław)
• 10th place – Barbara Brzegowa (Wrocław University of Environmental and Life Sciences)
• 11th place – Adrianna Kruk (Wrocław University of Environmental and Life Sciences)
• 12th place – Barbara Ptak (University of Wrocław)