Archive for the ‘ visualization ’ Category

3 days of #ehec on twitter – a visualization

There are 2 really recommendable tools for social network analysis out there: NodeXL and Gephi. Both are very powerful in conducting social network analysis and bring some handy features along. What I really like about NodeXL is the crawling feature with whom you can easily get some dynamic data out of Twitter. On the other hand there is Gephi with a cool yet not so user-friendly timeline feature.

Impressed what André Panisson did here to visualize the Egyptian revolution on Twitter, I tried my own.
At the moment we have an ongoing ehec-bacteria epidemic in Germany. Quickly the hashtag #ehec arose an new information spread quickly via Twitter. For 3 days (June 11th-13th 2011) I collected all the Tweets containing “#ehec” as well as retweets or mentions of other users with the help of NodeXL, then imported the data into Gephi to create this video:

ehec on twitter video

Bigger nodes mean bigger centrality. The users twitgrap and menno020 stick out since a tweet mentioning both was retweetet by sourrounding users. As you can see, over the days some clusters are formed, some users gain more attention than others, information is spread. Yet most are retweeted/mentioned only 1 or 2 times.

In the context of #ehec, the following hashtags were used most:

  1.  #ecoli (14)
  2. #EHEC-Tote (13)
  3. #Bakterien #Darminfektion (8)
  4. #Ecoli (8)
  5. #Focus #StN #BILD (8)
  6. #Pfingsten (8)
  7. #EHEC-Erregers #Monsanto (6)
  8. #DPA  #Ahja (6)
  9. #Google  #rassistisch (6)
  10. #STEC (6)

So this short list gives a quick overview in which context #ehec was discussed most: besides the bacteria itself, also the mass media more precisely the national press (Focus, STN, Bild) as well as Monsanto are mentioned. The latter in a rather conspiracy theoretical context. In the next days and weeks i’ll get deeper into Gephi which seems to be really really interessting, surrounded by a very active community. I’m looking forward to the upcoming versions!

Advertisements

kuler – finding matching colors

Ever wondered which color to take for an own graphic or presentation? Adobe’s kuler provides a quick and easy way to find matching colors.

Creating a heatmap with Photoshop (for NodeXL)

Hi everybody,

i spent the past hours searching for a nice way to create a heatmap for NodeXL. The aim was to provide something like this for any given NodeXL network. By the way, this technique presented here is not limited to Node XL networks. You can create a heatmap for almost everything using Photoshop (just skip step 1). If you don’t want to go through the whole process, you can download the PSD-file used here.

Edit: At the moment, all the heatmap-ish look has to be applied manually. If you have a good idea how to copy-paste a network and get to the same results, please comment.

1st step: Exporting your networks

In NodeXL, export your image as png. Preferably, you take a black background with white node and edge color. By then, you should have something like this:

2nd step: Preparing Photoshop

Open the image in Photoshop, and create 2 more layers. One background-layer and one to paint on later. You can fill the background with any violet-blue color. In this case it’s RGB 77/4/120.

3rd step:  Applying layer style effects on layer 2

In step 4 i want to paint the heatmap on layer 2. Therefore some layer style effects are needed. Select layer 2, click on the layer-style button (“fx”) and activate “Outer glow”. Set

  • “Blend mode” to “negative multiply”
  • opacity to 90%
  • spread to 10%
  • size to 20 px
  • range to 100%
  • and the rest to 0.

Next, we need that rainbow gradient. Just set the colors like shown beneath. It’s:

  • white, red (255/0/0), yellow (255/253/0), green (30/255/0) and blue (10/0/178)
  • important: set the upper stops to the following opacity values: black=100%, darkgrey= 80%, light grey = 55%)

4th step: Adjusting the brush

Select a brush and do some settings. In this case, i activated shape dynamics and smoothing. Set the opacity of the brush to 1% (as color, choose the background color), and to a size which seems appropriate for your needs (here: 5pxls). Go back to your layer 2 and now paint over your network. You should get something like this:

Or without the network-layer 1:

Conclusion

The method presented here shows how to manually create a heatmap of any given NodeXL network with Photoshop. Since this is not an automated process, you can actually create a heatmap out of everything (if you have enough time and creativity). For example an “infrared-image” of a human person:

If you have any suggestions how to simplify this process to make it more scaleable, please write a comment! Have fun.