Today’s post isn’t directly related to MicroStrategy, though it does have it’s roots there. At the last two MicroStrategy World conferences, one of my favorite presentations has been the ones by LinkedIn. They have some very unique data and challenges, and hearing the story of how they overcome them is very interesting. They utilize a wide range of tools to run their business, including MicroStrategy of course, but one passing mention on a slide caught my eye this year. They mentioned that for visualizations, one of the products they used it called Gephi (you’ll recognize it as the technology that powers InMaps). It’s an open source visualization package primarily for Network Visualizations. This is something I’ve been on the hunt for the last few weeks, so I was excited to check it out. Boy, did it blow me away!
Their tagline is, “Like Photoshop for graphs” which is a succinct way to put it. The tool is very, very easy to use and yet can produce some very powerful images. At this point, I’d normally dive into a quick tutorial and list of steps, but the Quick Start guides they have available are so good, that there’s not much I could add.
OK, there is one thing I could add. The Quick Start guide starts with importing a prebuilt data file, which is easy for the sample, but doesn’t really help you when you’re ready to explore your own data. To do that, you’ll need to create two CSV files: nodes.csv and edges.csv. nodes.csv should contain a unique list of all of the nodes in your network in two columns: Id and Label (the names are important). Your edges.csv file should contain three columns: Source, Target, Weight (again, the names are important). Source and Target should match the Ids from the nodes.csv file, and represent the links between two nodes. The Weight is if you want to differentiate lines by color or size. Import the nodes.csv file first, and the edges.csv file second and you’re ready to rock! Follow the rest of the Quick Start guide to build some cool graphs with your data.
One other quick tip: You pan around the screen with a Right-Click Mouse Drag, not a Left-Click.
After an hour of playing with the tool, I was able to generate this pretty picture: