Summer Research Week 4

Big changes! Lots of progress!

Version 2.0

In an effort to save massive amounts of time, I developed a new way to display all 58 graphs on 1 single web page, saving me the hassle of copying and pasting any new code that I may have to 57 other HTML files. This took a while to do, and it’s definitely not the way I envisioned it to look, but it gets the job done and that’s all that matters as of right now. Utilizing HTML form submissions, it allows a user to submit the name of the graph in which they want to see visualized. Using this submission, the ID of the selection is then passed through the Javascript code that was already written out and used when calling the file location. This way depending on what the user has submitted in the “form” the Javascript code will call a different file to read the network information off from.

1014adeaf082aa442ddf0d44c70d6285
Not only does this eliminate the need to go back to the index page every time the user wants to navigate the database, but it will save me so much time when writing code pertaining to a graph as all graphs will now be running on the same set of code.

 

Color Changes

As you might have noticed above, I decided to change up the color scheme a bit to help emphasize the data a bit more. Now you can clearly see which nodes have a stronger connection than other nodes. Looking at the GIF above, you will notice a chunk of the nodes connected with a very vivid bright green color while the rest are connected with red/brownish links. These brightly colored links have a stronger connection with one another than the darker colored links. This was true even before the color changes, but they are a bit more emphasized now. Now, in order to attain a bright green color, the link between 2 nodes must be at least 20 or above (changed from the original 3 or above). Because of this shift, you will now notice the presence of a brown color. Because of the way the color scale works, this brown color is represented as roughly the number 10. Meaning that any links colored red are <10, anything brown is ≥10 but <20, and anything green is ≥20. I tried experimenting with different color schemes for fun such as tri-tones and even the color spectrum, but they all seemed too distracting.

color_spectrum_attempt
For fun, I tried doing the color spectrum, going from Red -> Purple
Link Values

Colors are nice and all, but what if we want to know the exact values between the nodes? Well, now it is possible! Before in order to know the value of a link you either had to guess using thickness and color of the line as a reference or open the JSON files are parse the information yourself. Now users can simply hover over a link and see for themselves! Yay! b16592b77a8653d239cd27eede255a14

Accurate Numbers

I noticed that almost all the graphs had pretty low densities compared to what they should have. I quickly realized that the TSP Study nodes were what was causing this inaccurate calculation. After fiddling around with the parser code on Eclipse Java, I managed to get rid of all TSP Study’s, and all Nurses/Doctors involved in the Facebook groups. Every graph shot up significantly in terms of density, and now all the graphs display a more accurate value for their density. Here is a side-by-side comparison of what it used to be vs. what it is now.

graph_with_TSP_Studys
Before

 

graph_without_TSP_Studys
After
Next Week

I’m currently in the midst of developing a way to automatically assign a node or nodes to be the “central node”, meaning the node with the most connections to other nodes. I have all the algorithms and theories on paper, but I’m contemplating going over all the previous code I have written in the parser and thoroughly cleaning it up. Introducing a node class would be nice, and even the thought of a network class which holds all these nodes is also a possibility. I have a function all written out to calculate a node’s centrality, but I have yet to implement it (something to start working on as soon as I’m done writing this article 🙂 ). This week was very exciting and I look forward to what’s to come next week. Stay beautiful everybody ❤

~Abir

 

 

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s