Case Study in Travel Business - Understanding agent connections using NetworkX

Network analysis is getting more and more attention in Business Intelligence, people hope to get information out of the structure of an organization or a communication network. In this talk, we use the hotel room search requests from travel agents, including online public website, B2C, B2B and B2B2C, to build a relational network among them. By using this network as an example, we demonstrate how insights can be extract by studying network properties.

In the first half of the talk, we will explain how the network is built using NetworkX, an open-source python library that is designed for the creation, manipulation, and study of the structure, dynamics, and functions of complex networks. When 2 agents are making the same search at the same time , a link ( or an “edge” in network analysts terms) is made pointing form the initial searcher to the subsequent searcher. Using a list of these searches, a directed graph is built. We will also demonstrate how to pick the biggest connected component out form the graph. In the second half, with the graphs created, we show how different functions of NetworkX can be used to study the graphs. By compare the graph properties of our graph to the other popular network graphs, we can get the insight of how the network was created. Also by studying the graphs, we can understand the behavior of the agents and can even figure out which agents are acting as main hubs in the network.

This talk is for people who are interested in network analysis and would like to see how it can be used in a business case. Audiences with any level of python experience can learn some basic concept of network analysis work and how it can be applied to provide business insights.