The town hall of Paris has set up a wifi system to allow parisians and tourists free and high-speed wifi access to the internet throughout the city. This internet access is distributed through hotspot terminals distributed throughout the city.
The principle is as follows: anyone wishing to have internet access can connect to the internet network distributed by the city of Paris, by registering and providing their identity. The user then has access to 2 hours of connection. (Note that there is no connection limit. Once his 2H package has been used, one can reconnect to have 2 new hours).
The number of hotspots, their density and their positioning is different for each of the 20 arrondissements in the city of Paris. Moreover the activity is different according to the periods of the year and the days of the week. Thus, the number of users per day is a variable data just like the total internet consumption. (Quantify by the in and out flow of data of all users).
These hotspots are therefore subject to uneven daily use depending on the boroughs, and predicting the use of wifi in the different areas of the city will allow better management of this device. (To perform maintenance, to invest in areas where use is important). And thus will enable better service for customers.
This is therefore what this challenge proposes, we want to predict internet consumption, ie the number of gigabytes entering and leaving according to the districts, taking into account past data. We'll elaborate on the framing of this prediction, but first let's look at what the data we have looks like.
This project have been done in collaboration by Aurélien Renault, Ayoub Tabaai, César Leblanc, Ikhlass Yaya-Oyé, Oumaima Bouther and Victor Clermont.