Introduction

Airports connect one place to another, and the air networks make traveling more accessible and more comfortable. It’s a common thing to travel by air.While the number of public-use airports in the United States has fallen since 1990, the number of private-use airports has increased. In 2019, there were 5,080 public airports in the U.S., a slight decrease from the 5,145 public airports operating in 2014. Even back in 1997, the US air network was one of the world’s busiest transportation systems.

In this project, I focus on visualizing the 1997 US Air networks and understanding how much people rely on traveling by air in 1997.

Process

For this project, I collected from Gephi-Wiki. It is one of those websites which provided meaningful and exciting data. I selected US Air97: North American Transportation Atlas Data (NORAD). Gephi is software allows using open source network analysis and visualization by importing nodes and edges table and complies them in a network. Users can run different statistics on the network after that.

Since the data file I had was .gexf, it could be imported into Gephi directly. It supported to import several kinds of files, and allows users to edit the data within the software. Therefore, users could explore different kinds layouts.

This data was based on two dimensions: labels(the names of the Airport) and ld(parameters of airports). After importing the data in Gephi, I ran the statistics, and started my visualization by trying out different layouts that could suit my data. I also explored the filter option to filter out data with modularity and degrees.

My inspiration was from this article, as it is common to have a world airline system presented in network visualization. However, the dataset I had was simple. The result of the visualization could only be delivered in one direction.

Results & discussion

The results mainly reflect the connection among one airport and others. By adjusting the layout of Fruchterman Reingold, the final results looked very packed and highly-connected. It leads less readily to differentiate all the data, as there were around 300 different labels only by color-coding（Figure1）.

Then I switch layout mode to ForceAtlas 2（Figure2）. Then I switch layout mode to ForceAtlas 2. It looked less complicated, the relationship among data does seem more organized, but it stresses less on the connection between two different airports.

Unlike what I did in Figure 1 and Figure 2, I color-coded with degrees other than modularity.

Reflection

Overall, Gephi is a helpful tool to realize data visualization. Its user inference has relatively straightforward guidelines to follow and tutorial documents that helped present the visualization positively. However, when I look back at this entire process, I was surprised to find that the most challenging part was the raw data. As my dataset only includes two dimensions, the results have become very straightforward, making it hard to explore more possibilities.

As for my future steps, I would consider working on a visualization result with more creative content and rich dimensions. Besides, I think I should be prepared and ready for research for adequate datasets and practice of using tools.

Sources

Gephi Wiki https://github.com/gephi/gephi/wiki/Datasets

Networks http://vlado.fmf.uni-lj.si/pub/networks/pajek/data/gphs.htm