Data visualization: Link Analysis
Lately the concept of ‘Link Analysis’ gains more attraction. We see these analyses on more places then ever. And that is because there is more and more data that indicates what relates to what or who’s related to whom.
We probably all recognize a visualization like this one:
Facebook, LinkedIn, all those big social media sources have lots of interesting links in their data. Who connects with whom. But next to that we could also think of enterprise e-mail or connections in nature regarding the food chain or the relationships among those ecosystems. Lots and lots of connections and relationships there…
I recently saw a few interesting pieces of information regarding this type of analysis, to name a few:
- Market basket analysis (http://www.analyticsvidhya.com/blog/2014/08/visualizing-market-basket-analysis/)
- The no longer supported ‘InMaps’ from LinkedIn (https://www.youtube.com/watch?v=PC99Nw2JX8w)
- A Wolf’s Role in the Ecosystem of Yellowstone (http://www.neo4j.org/graphgist?0ac320c799ce55089377)
Now I think these pieces are very interesting. But to be able to read those complex visualizations you have to dig in. The most common problem, and certainly my challenge with it, is that these visualizations are presented without axes. So for me it is hard to figure out what and where the references are. But maybe that is my challenge only? 😉
When I looked at the data more closely I knew that the data in itself was very simple. The values we have are in its most basic format:
Source;Destination;Value 862;841;144 40;460;208 887;960;176 322;12;180 733;938;182 420;856;58 870;416;121 289;861;123 556;267;144
But next to this example I realize that this set doesn’t show multiple intersections that could also be possible. But for the sake of simplicity, this is going to be the sample data that I’m going to use.
My question at this point: “How could I do this (hopefully) more simple with a native QlikView chart?”
“It should be possible to do these kind of analyses with a simple line chart, I guessed.” I had seen visualizations like this before, for example:
So there should be a method in QlikView for this sample data that I showed earlier.
It took me a couple of hours to figure it out, but here is the result of the visualization that I developed with QlikView:
To be able to develop this chart I had to overcome a few challenges:
- The height of the arc had to represent the ‘Value’ field in our sample data;
- Drawing a line in a ‘line chart’ for the matches wasn’t that simple. The arc that is created here needs to be lifted on the average of the two link values;
- The weight of the line had to give an indication of the relative value of that connection against the total of all connections;
- To make it even more intuitive I wanted the line color to stand out for those connections that are very relevant in terms of their contribution to the total.
For this I had to figure out some transformations of the original sample data to help me with the challenges above.
Thinking of the chart at this point I figure that the X-axis could also help to maybe indicate a few other relevant measures as well, maybe this reference could also help me to show:
- How long certain connections exist in time;
- At what age these people/products/etc are themselves;
- What the total value for that individual link is or a classification bucket for that matter;
- Or maybe the physical distance between the two connections.
Any way, this was a bit of a brainteaser or some random ideas for more or better insights that could come with a visualization as this one.
The only thing left here for me to do is to share this app with you guys so you can have a look at it. I hope this all makes sense, if not… Just punch in some questions in the comments.
The app can be found here.
Happy designing with your next applications. Stay tuned…