PageRank

Purpose

The project I did for my Information Retrieval course. This project is solely for sharing my implementation, not for further development.

Description

Implementation of PageRank algorithm using the flights data from http://www.openflights.org. The algorithm calculates the PageRank value for each airport in the input data and returns the TOP-15 most popular airports.

Data

The input file data/routes.csv contains 59036 routes between 3209 airports on 531 airlines spanning the globe. The file contains two data fields: 3 letter IATA code for an origin airport and for a destination airport. Routes are directional: if an airline operates services from A to B and from B to A, both A - B and B - A are listed separately.

Stack

• Language : Python 2.7
• Libraries : NetworkX, Pandas, Datetime, CSV, Operator

How to run

• Run src/PageRank.py
• Specify the full path to the input file routes.csv

Notes

• For the purpose of PageRank performance optimization, the src/GraphConstruction.py file pre-processes the input data on routes by first building a weighted graph of routes and further representing it as a nested dictionary.
• You can modify the damping factor and the number of iterations in the following line of src/PageRank.py : airports_page_rank = page_rank(0.85, 100)
• For a full description of data pre-processing and PageRank algorithm check my report in the docs folder.