| KEYWORD |
Dynamic zoning algorithm based on mobility patterns estimation
keywords DATA MINING, NONLINEAR ANALYSIS, COMPLEX NETWORK, INTELLIGENT TRANSPORT SYSTEMS, MODELLING AND EXPERIMENTAL TESTS, STATISTICAL DATA ANALYSIS
Reference persons CRISTINA PRONELLO
Research Groups Transport Research for Innovation and Sustainability (TRIS)
Thesis type DATA ANALYSIS AND MODELING, DATA MINING, EXPERIMENTAL AND MODELING
Description The first step of transport supply models is the zoning of the territory to allow to assign the user trips from one point to another one that are the centroids, that are the barycentre of the zones. Of course, this is a simplification of the model because it is not true that all people in a zone start/end their trip in the barycentre, but they could start or arrive in a point close to the border of the zone, quite far from the barycentre. In this case, the route from origin to destination is not the right one, causing a big mistake in the forecast made by the analyst; the spatial distribution of the mobility patterns is thus inaccurate, and it is impossible design the transport systems that properly fit the user needs.
The division of the territory in zones is made according to a few rules but it is not deterministic and suffer of a certain subjectivity. So, instead of just assigning trips to the barycentre of existing zones, we would like to make a dynamic zoning. This implies to draw the zone to minimise the error in assigning the trips (start and end point) to the centroids and redefine the zones.
Students will be provided some mobility data (in the form of user trips) that were previously collected through a mobile application. In case the app is not available, students would need to simulate some trip data.
The students will need to use these data and the Zoning data available for the city of Turin and try to identify algorithms that will enable the re-definition of the boundaries of the zones based on user trip. For example, wherever there are more trips starting in a small area may need to be a separate zone even if that area crosses two existing static zones. Students can also attempt to understand the feasibility of having two different zonings for origins and destinations. Any doubts about the transport concepts and reasoning behind what needs to be considered for designing the algorithm will be explained in detail during the course of the thesis.
The expected outcome will be a system that dynamically changes the zone structure based on available mobility data and pre-existing zone structure.
Required skills Preferably, General programming knowledge and algorithm development, data collection and analysis. A knowledge of transport systems is also fundamental to better exploit the potentiality of the ICT. Python is the recommended language, but students are free to use whichever language they feel comfortable with.
Deadline 05/11/2025
PROPONI LA TUA CANDIDATURA