Missing data. In some circumstances, when indispensible data was missing, the entry was not included in the database —e.g. Building ID, X36 NPV List. In the case of Network Proximity, data was imputed using the smallest approximated value calculated by adding the distance to the closest buildings that had values and its Network Proximity.
//TODO Details of DB building process We have several end points on the server to access the data. These are: /market_profits: /building_profits:
These work encompasses four visualizations:- Overview Choropleth Map [1]. The user can select [2] a market and view the total profit of sites not on Zayo Network for that Market. In addition, the user can view a summary [3] other aggregated values for that market as a list.
- Most Profitable Buildings Map. Overlaid on a google map of the selected market in the Choropleth Map, the user can view the buildings in that city that are not on Zayo Network. The user can filter [4] the buildings to view according to different criteria.
- Attributes Histogram. Based on the data reduction from Most Profitable Buildings Map, the user can select an attribute from a menu to visualize its distribution [5] as a histogram.
The color saturation is used here as a magnitud[7] channel to represents the total profit per Market. The total profit is calculated by aggregating the values of attribute X36 NPV List for all of the sites that are not on Zayo Network, but located within the specific market. We used an intuitive direct proportional relationship, that is, the higher the saturation, the higher is the total profit. Another marker such as position on a common scale would be more effective in the representation. However, to keep an efficient and simple design with the underlying map, the actual number is render with a mouse-over action.
Typically in choropleth maps, normalized data is preferred to raw data to overcome a common interpretation bias –e.g. when presenting relations that include population.[8] However, we chose total profit –as opposed to a profit density- based on the premise that X36 NPV List reflects already all overhead related to management of the sites –as understood from our discussion in class with a Zayo representative. In a given market, if the premise is correct, more accounts for the same profit would be beneficial for the sustainability of the business. For example, in one market two accounts render the same profit as 20 accounts in another market. If one account leaves, there is a higher impact in one of the markets than in the other. Independent of the business strategy for client proliferation, the attribute total profit is relevant to towards the main goal.
More information is provided to the user for any selected market by a mouse-click (circles are playfully highlighted by changing size). The overview list contains aggregated values of the number of buildings, the number of Sites, the number of Accounts, the top of Product Group, the average Network Proximity, the average Estimated Build Cost, the Industry type, and the Vertical.
-
Most Profitable Buildings Map. This is a simple visualization of the geographical location of the buildings in a selected market. The selection occurs in the previous visualization, but this map is rendered on a different frame (div).
-
Via a drop down menu the user can filter the view of the buildings that satisfy different criteria –e.g. top ten, top x percent. We selected a cleaner dropdown menu in opposed to the more cluttering option of labeled buttons.
-
Attributes Histogram. As the name describes, this visualization uses histograms to show the distribution[9] of the values for an user selected attribute. The data used for the histogram has been reduced in the previous visualization Most Profitable Buildings Map. Once more, the selection of the attribute happens via a dropdown menu.
-
Michael Iuzzolino: Front-end development, D3 info visualizations and interactive interface design, Google Maps API integration.
-
Shirly Montero Quesada: Readme, interface conceptual design and review
-
Athithyaa Panchapakesan Rajeswari: Database/Server/Designs on Server, interface design, visual design and styling
-
Santhanakrishnan Ramani: Database/Server/Designs on Server, interface design, visual design and styling
-
Nicholas Schardt: Proof read