Due: Tuesday, 3/28. Submit a hastac blog post by classtime (we will discuss them in class) that shares images of your networks created in Cytoscape (to the best of your ability). Here are options/directions for how to share Cytoscape networks. Here is all of Cytoscape’s 3.4.0 User Manual for troubleshooting and further work. (You will also find answers to your questions in Google results.)
What question are we trying to answer and what is our data set? In other words: why use Cytoscape here to create a network visualization to analyze?
I want to visualize the distribution of race and ethnicity among the deceased that law enforcement agencies killed in 2016. I also wonder which law enforcement agencies targeted which demographics of victims. Using this data, we could then focus activist watchdog efforts on those agencies with the most fatalities for one or more groups, and we could also alert populations that are most in danger. Where will I get this kind of data? We have already looked at it in the project called The Counted. Data in “The Counted” is extensive, so it’s difficult to visualize in tabular format, especially if we want to compare two sets of data. In general, can we see who law enforcement agencies have targeted in 2016? And what do we learn when we compare this to 2015 data?
In class we will visualize the 2016 data. On your own, you will do the 2015 data set. Your practicum will record and report the findings of your comparison between 2015 and 2016 network visualizations, with screenshots or other kinds of sharing (see Cytoscape sharing options linked in intro). Remember: the visualizations you produce are not your analysis. Your syntheses, evaluation, and the interpretive conclusions that you draw from your visualizations are your analysis. Your job is to use what we’ve learned from James Adams, Franco Moretti, and others this term to offer an interpretation of this data that is critical analysis. Your blog post will indicate to me if you know what critical analysis is and how to perform it. If you are unclear on how to do this, please ask me. You should also record what you cannot learn from your network analysis; what can it not tell you? Where does it fail?
Last: if you spend your allotted practicum work time simply struggling to create your network and “fighting” with Cytoscape, analyze that process and what you can learn from error messages or the struggle to curate your data and make the application work. It’s okay if it doesn’t work perfectly for you — this has been our refrain with practicum exercises this term.
Directions we will follow together in class (you have these now to replicate the process on your own with the 2015 data set).
- Download Cytoscape from http://www.cytoscape.org/
- Open it up. Ignore the dialogue box that pops up – close it. Now, minimize Cytoscape and you will focus on creating the dataset to import into the application.
- Go the The Counted data page and download the 2016 data. You will find files for both 2015 and 2016 data. Open the 2016 file. Call it the-counted-2016_raw.csv and save it to your desktop or to a memorable file. This way, you have an “unadulterated” raw dataset to consult that is separate from the dataset you will curate for Cytoscape.
- To create a network visualization, you need a data set with, at minimum, three columns.
We are going to create such a dataset out of the 2016 data from The Counted by altering our raw file.
Go into the csv file and click “save as” and call it “the-counted-2016_network.xlsx” (Note: .csv files will not upload into Cytoscape. You must save as .xlsx or you will get an error message.)
- In your “network” file, delete all the columns EXCEPT for
THEN you are going to reorder them in your spreadsheet so that column A is “lawenforcementagency”, column B is “classification”, and column C is “Raceethnicity”.
- lawenforcementagency is your Source (Column A)
- Classification is your Interaction type (Column B)
- Raceethnicity is your Target (Column C)
- In Cytoscape, Select File > Import > File > Network > the-counted-2016_network.xlsx. You will then get a dialogue box. In this dialogue box, you need to set your Source, Interaction Type, and Target:
- Click on the column heading “lawenforcementagency” and then select the green circle to make it your “Source Node”
- Click on the column heading “classification” and then select the purple arrow to make it your “Interaction Type”
- Click on the column heading “raceethnicity” and then select the orange bulls-eye to make it your “Target Node”
- Click OK. You should see a network visualization (probably a messy one) now.
- Use the magnifying glass icons to zoom in and out on your visualization. Notice that the labels for your targets and source do not show up until you zoom waaaaay in. And they are confusing to read. We can’t yet make heads or tails of this network. We are going to change the appearance of the source and target nodes in order to make this visualization more readable.
- In your control panel top menu, click on “Style”. In the bottom menu, select “Node”.
- There are too many law enforcement agencies to read each dot. So let’s hide those names. You can see them when you click on them. Select “Label” and “Discrete Mapping” so that you select which labels show and which do not. I select those labels for race/ethnicity from the list: Arab American, Black, Hispanic/Latino, Native American, Unknown, White
- However, the Source and Target nodes still look the same. Let’s change their colors and shapes to differentiate. Still in the “Style” control panel, find “Color”. Change the color of the race/ethnicity targets and make it different from the source nodes. I turn the source nodes red and the targets a light blue. Then go ahead and make the targets much bigger than the source. You can do this by checking the box “Lock node width and height”. Select discrete mapping again and then give the source nodes a higher value. I set the targets at 100 and the source nodes at 20.