ATTENTION/WARNING - NE PAS DÉPOSER ICI/DO NOT SUBMIT HERE

Ceci est la version de TEST de DIAL.mem. Veuillez ne pas soumettre votre mémoire sur ce site mais bien à l'URL suivante: 'https://thesis.dial.uclouvain.be'.
This is the TEST version of DIAL.mem. Please use the following URL to submit your master thesis: 'https://thesis.dial.uclouvain.be'.
 

Amélioration des zones d'adresses chez Bpost

Teugels, Simon
(2019)

Files

Teugels_52751400_2019.pdf
  • Open access
  • Adobe PDF
  • 2.37 MB
Download
Teugels_52751400_2019_Appendix1.zip
  • Closed access
  • Unknown
  • 223.06 KB

Details

Supervisors
Nijssen, Siegfried
Faculty
Ecole polytechnique de Louvain
Degree label
Master [120] : ingénieur civil en informatique
Abstract
enIn a context of growing mistrust of large companies and their data management, Bpost wishes to combine privacy and performance in order to provide its customers with a quality flyer distribution service. In this master thesis, we were asked to find a solution to count the addresses to distribute flyers to. In order to make this possible, it was necessary for Bpost to propose a relatively simple method to select the geographical area in which the flyers will be distributed. The main criterion to be met was not to allow users to discover an area with less than 80 addresses. This constraint is based on Bpost's reading and interpretation of the GPDR. We have also had to comply with other efficiency constraints regarding the distribution centres and activities segments to which the selected addresses belong in order to facilitate the procedures related to this service. The plan was first covered with polygons using a Voronoi diagram using the addresses as germs. We then merged these polygons until we reached the threshold of 80 addresses previously set, while respecting the criteria imposed. The algorithm we have built, coded in Python, is inspired by Casper's algorithm and its pyramidal structure to form several levels as well as its bottom-up operation for the actual address selection. As the use of these levels is subject to potential attacks, we also considered the possibility of using only the last level. The results we have obtained are satisfactory in both cases. Of course, the accuracy deteriorates somewhat when we use only the last level. However, we can still consider using this version if the risk of attack is considered too high. The objective of providing Bpost with a solution that respects anonymity and is as precise as possible in order to count the number of addresses in a given area around a point is achieved.