Master theses

Predict how busy a certain part of the city is with Open Data

Keywords: AI, open data, artificial intelligence, deep learning, tensorflow , smart city

Supervision: Pieter Colpaert Ruben Verborgh

Students: max 1

Cities want to know how busy certain parts of the city are. This is perceived by citizens as if the city that wants to be “smart”, is now going to track them (e.g., check out this article about the city of Kortrijk in De Standaard). The goal behind the idea is however straightforward: if local businesses are saying they want to leave a city because there are fewer people, we need data to fact-check whether this is true (e.g., check out this article about the Ghent circulation plan at VRT NWS).

Thus, we want to give, in all transparency, a good view of how people are moving in a city (real-time and in the past), without violating the privacy of individuals. We think different non-personal data can give a good indication of how busy a certain part of the city is: for example, number of people connected to public wifi, how full parkings are, amount of trash being thrown away in smart garbage bins, bike counters, events that take place, bus and train occupation, etc. Can we reliably predict how busy a part of a city in Flanders will be based on this kind of data?

As part of the Smart Flanders programme, we are publishing open datasets that may contribute to such a prediction. It is your task to come up with a method to calculate such a prediction and give a nice overview for the 13 biggest cities in Flanders.