Home

Awesome

Finding Continents from a Flight Routes Network

Omar Boujdaria, Franck Dessimoz, Arnaud Duvieusart and Adrien Vandenbroucque

This repository contains the code for the Project of EPFL's Network Tour of Data Science course. The goal of this project is to perform community detection on a network of flight routes, with the goal of identifying the continents.

Packages required

In order to run the code properly, you will need the following packages:

The structure of the files is the following:

Data

The datasets we used can be found on https://openflights.org/data.html. We used the files routes.dat and airports.dat.

The file routes.data contains informations about the flight routes, more precisely it contains the following features:

The file airports.datcontains informations about the airports, more precisely it contains the following features:

We also included the cleaned files that we further used in the project, called routes_clean.dat and airports_clean.dat

About the notebook

The first part of the notebook is dedicated to preprocessing and cleaning of the graph.

In the second part, the main work of the project is shown, that is first a review of the main propoerties of the graph, followed by some visualization, and the community detection part. In this part, we present different algorithms, and show the resulting communities. We also measure the quality of the partition by measuring modularity and coverage.