"I have neither given nor received aid while working on this assignment. I have completed the graded portion BEFORE looking at anyone else's work on this assignment. Signed Danny Nguyen"
I would like to create a visualization of the backbone fiber segments of UNC’s fiber network through the underground ductbank system. I will use data from UNC’s Fiber Manager application within the ArcGIS environment to create a better visual based on the concepts we have learned in class. I plan on creating a geographic visual which will overlay the backbone fiber segments onto UNC’s campus map. While the original visualization will be created using ArcGIS, I believe that this visualization will only be useful to others as a PDF, since most users do not have the ArcGIS tool. As a PDF, users can interact with the visual by zooming in/out of points of interests and can print out the document on plotter-sized paper to view the map.
This topic is interesting to me, because it is a practical application of the concepts in class toward a real need in my day job. While a current visualization for all the fiber segments on campus exists, there is not a specific visual created for only the backbone fiber segments. Additionally, the layout and color scheme of the current visual could be improved by our visualization techniques learned in class. UNC has had 3 major phases of fiber upgrades in its history, but little is known about the aggregate capabilities of the fiber backbone network created by those three phases.
For this exercise, backbone fiber segments will be defined as any fiber segment of 144 count or more of fiber strands. The original fiber data is part of a large database of all of the fiber segments on campus, but for this assignment, I would like to focus on only the backbone fibers, since these segments are of greatest concern to management while allocating fiber or discussing how much fiber exists. Backbone fiber segments feed other buildings on campus with fiber and typically connect back to the ITS Manning building.
The tasks that I believe need to be done are:
Query the database for only fiber segments that meet our criteria for backbone fiber segments
Draw fiber segment connections from origination to destination locations
Decide how much information to display about the campus (e.g. Use an aerial map? Use a simple campus map or a detailed campus map? Highlight certain buildings such as Tier 0 and Tier 1 buildings?)
For my midterm deliverable, I plan on creating an actual visualization and delivering a PDF version of the ArcGIS map of UNC’s campus with the backbone fiber segments represented geographically.
*******************************************************************************************
The Context of this visual is to display all the fiber segments on UNC-CH's campus that are considered backbone fiber - which was redefined as any fiber segments with 128 count of fiber strands or greater. The User for this visualization would be a senior administrator in ITS (i.e. ComTech Engineering Manager, Director, CTO, CIO, etc.) The Task that this visual is addressing is showing the backbone fiber segments of UNC's fiber network through the underground ductbank system as a geographic visual for strategy and planning purposes. This visual also allows the user to view the aggregate capabilities of the fiber network created by different phases of construction (Gen 1-3 designation).
Unfortunately, I could not export the database data from ArcGIS easily; however this data was created by inputting hundreds of tabs of Excel spreadsheets into the database (see above image for sample source of data). For my visual, I had to query the database of information to extract the pertinent data that I wanted to display and then manipulate the data to also derive the temporal information about different generations of fiber. The data types that were prevalent in this visual were ordinal (fiber strand count and Gen 1-3), temporal (Gen 1-3 dates), nominal (UNC Buildings and names), and cartographic (geographic map representation).
The visual was created using a combination of ArcGIS (with the ArcFM/Fiber Manager add-on) and Adobe Acrobat Pro X. The final visualization was represented as a Adobe PDF file, because PDF is a familiar format that is universal to different operating systems. While the visual would look better and be more interactive in its native form as an ArcGIS file, this program is not intuitive to use and others can't access the file without this large program. Additionally, I had to "learn" (through hours of trial and error and experimentation) how to use both ArcGIS and Adobe Acrobat Pro X during this midterm project to be able to create the final visualization.
In this attempt, I removed a lot of the distracting data and changed some of the color schemes to be more visually appealing. I created a symbol and left in the manhole locations for the ductbanks (without the name labels) because I thought that this information might be useful. I later decided that the color scheme and manholes needed to be changed/removed.
In this attempt, I queried the data to show all of the fiber segments that met my initial criteria of outside fiber segments that were greater than 144 count of fiber strands; however, this representation did not give me much detail. To be able to show more information and allow the user to interactively select different fiber strand counts or groups of fiber segments, I needed to create multiple separate layers and tease out the data. I changed a lot of the color schemes and used the Color Scheme Designer to determine good complementary colors to use instead of the original colors that were chosen. I removed a lot of the extraneous data that cluttered the visual (such as the Chapel Hill and Campus Base maps, Facilities Services underground structures and conduit systems, etc.). I added insets to the map for displaying parts of UNC-CH's campus that were far off campus and were not easily viewable as a whole map. I added a title, scale, legend, and revision note to the visual. I attempted to use a lot of Stephen Few's guiding principles for removing distractions from the visual to communicate an efficient story. I decided to draw a box around my legend to delineate it from the rest of the visual, because there was a lot of "noise" in the background with the streets drawn in. I decided not to label the Orange County streets because doing so created a big distraction since the street names were scattered throughout the map.
In this attempt, I created separate layers for the different fiber strand counts and color coded them with complementary colors from the Color Scheme Designer. I later decided that the color scheme that I chose was not very good because the colors did not tell a coherent story, even though they were complementary colors. In my final product I decided to use a gradually increasing color scheme that was a complement to the predominant electron orange ductbank color. I also attempted to change the size of the fiber strand line symbols to represent smaller vs larger fiber segments, but by doing this I could not easily see the fiber segments anymore, so I decided to keep all the fiber segments the same size and used the lighter to darker shades of blue to represent increasing fiber strand counts.
IMPORTANT: see PDF attachment of INLS 541 - Midterm Project (Post) v 5.pdf below for full deliverable to access interactive layers
In the final presentation, I made all of the changes that I wanted in ArcGIS and then exported that data into Adobe Acrobat X Pro. From here, I had to create/rename/merge various layers to create the sleek layer tool bar. The output from ArcGIS required a lot of manipulation to create the actual PDF that I am presenting to the class as a final product. I also made the decision in the final product to include fiber counts that contained 128 count fiber strands as well, since my definition of backbone fiber needed to be expanded to include the first generation fiber that was installed. In my PDF deliverable, a user can either show fiber strand count information and turn off the Gen 1-3 fiber information or they can turn off the fiber strand count information and turn on the Gen 1-3 data. If the user leaves both visible, then they will only be able to see the fiber count strand data as the front most data. The ability to turn on and off layers allows the user to see various fiber strands that travel along the same ductbank paths and may be obscured by other fiber segments. Gen 1 fiber were all the fiber segments that were 128 count and Gen 3 fiber were all the fiber segments that were 288 count. Any of the fiber segments in between those two were considered Gen 2 fibers, based on the phases of construction.
Review Requests: I'd like or reviewers to help suggest ways to XXXX.
Danny,
I’m really, really impressed with this visualization, which seeks to put usability at the center to improve an existing visualization that looks to be intended for engineers and specialists, but not lay people and decision-makers on campus. The original visualization and accompanying data certainly provided an interesting challenge, and I think your approach is an elegant solution to that challenge. You have done an excellent job cleaning up the original visualization of the noise without compromising any of the data. Your final visualization is crisp, clean and makes good use of white space (without having too much empty space). Breaking out the extended campus areas in smaller boxes is an effective use of your limited space.
I like your choice of the blue-green multihued color scheme (I’m actually having trouble determining if this is single- or multihued, given the difference in display between the wiki version and the PDF) for the fiber strands, with the complimentary orange for the duct banks. This helps the duct banks stand out better, while still allowing you to show differences between fiber types and generations.
Your use of overlaid map layers is especially effective in allowing users to see differences and progressions. That’s what allows you to tell the story of the growth in type and reach of fiber segments.
I really like your use of the Adobe PDF output, since it is very easy for most users to navigate. However, not everyone is aware of how interactive a PDF can be (I had never seen a PDF like this before!). I wonder if you could pull these visualization layers into a web interface with an even more self-evident and intuitive set of interactive filters -- check boxes, perhaps, attached to the legend. I’m imagining an interface not very different from a typical map interface in Tableau Public (though I wouldn’t necessarily suggest that particular tool for your visualization). It might also make the visualization easier to update and disseminate, and thus more dynamic in the long run. That is to say, is there a happy medium between the complexity of ArcGIS and the static nature of PDF (by “static” I mean hard to update)?
Additionally, though you have a very specific target audience in mind for this visualization, I nonetheless wonder if a little more context might be helpful. For instance, you could add an additional sheet in the PDF with instructions and context. Or if you were putting this visualization into a web interface, such context (narrative) could be included on an introductory page. Perhaps you would want to include some basic instructions for how to use/interpret the visualization? Not much would be required, but it might help non-experts tease out important information more efficiently.
Great job! I’m hoping to learn ArcGIS and I’m impressed and envious of your skills.
-Pam
This was a really impressive bit of information visualization. You elegantly solved a multitude of problems with a single interactive visualization. I am most impressed by your layering of data by generation, fiber count, duct bank, and building infrastructure. This allows the user to parse the data in a way that makes sense to them. You have really moved away from presenting the user with a simple visualization, and have presented the user with tool by which they can analyze the UNC fiber backbone in a method of their choosing.
The only place I would really suggest an improvement is in your color choice. Various shades of green may have been a poor choice for displaying generations due to congenital color blindness. Perhaps you could make use of some of the color testing tools we have learned about to make sure that your visualization could be understand by the maximum number of people. Additionally, and this may simply be a personal problem, I have trouble studying visualizations with pure white backgrounds as they strain my eyes. Perhaps you could make the background of the visualization a neutral gray so that it is a bit easier on the eyes?
Overall, I have very little to add. It appears that you very neatly accomplished your goal. I hope that this visualization is appreciated by your place of employment.
Very nice project Danny. I like that you took a challenging work problem and tried to address it in a meaningful way. Good CUT-DDV discussion, and good choices of visualizations. I liked the iterations of visualizations. I was particularly impressed by your use of layers in the PDF view as a way to deliver a visualization through a standard (PDF) that allow just the desired interactive control needed (i.e. turn off and of layers). The one suggestion I have is to consider whether there are classifications/type of groups of network cable so that you could group these instead of having only interactions with all the individual layers (i.e. it makes it more cumbersome to see differences and interact with all the individual layers).