Introduction

For our lab on CartoDB, I decided to use a dataset, the Museum Universe Data File (MUDF) that I was already familiar with (from our Tableau lab, 2014-Q3). This dataset comes from the Institute of Museum and Library Services (IMLS) and includes information on over 35,000 museums in the United States. Initially, I wanted to see how many of each type (or discipline as it is referred to in the dataset) were in New York City (including all five boroughs) and more specifically where they were located. The dataset includes information on revenue for many of the institutions, but not all. Of the ones in NYC, I asked:

• How many of each type (or discipline) are located in NYC area? Is there any relation to the type and the location?
• Does location have anything to do with an institution’s revenue? Where are the highest and lowest revenue institutions? Are the in the same areas or different areas?

Inspirations

I was curious to see visualizations that were art-related as well as visualizations that had location points. I found two visualizations related to art in NYC here and here that look like they are for a class called Art & Politics in the City at NYU. Neither of these visualizations is all that successful though in my mind. Not particularly easy to spot information – I could not locate any works by San Telmo in the 2015 Street Art map (not wise to use the same base map color and point color). Both include photos in the info window which I like, but it would be better if there was some sort of  categorization or filtering to make it more engaging- even if it were simply- stencil vs painted, artworks on the sidewalk vs artworks on objects, or on buildings,…

I started looking outside of art-related and found one related to alternative fuel, which I ultimately found to be a more successful map of points. I liked the combination of different icons shapes along with colors for each type of fuel.

It is a little overwhelming at first glance, but ended up being fairly easy to navigate and zoom into areas that were dense and “un-stack” some of the icons. Some of the stacked icons seemed to be the same business though- perhaps they could have consolidated some of these for an overall easier view.

Visually this map of NYC Street Trees, created by Jill Hubley, really intrigued me and I found successful for it’s overall look. I wish that it offered more information about the trees and their canopies and maybe their allergenic properties like this viz. The filter option is a major plus for the NYC Street Trees viz. I wish I could combine the information in these two visualizations.

Below is side by side screenshot using the filter option. The image on the left is filtered for “Callery Pear” and the one on the right is for “Birch” trees. If you have an allergy to Birch, not so bad in NYC, but Callery Pear- not so great.

Materials 

• Excel
• CartoDB
• MUDF Dataset, IMLS 2014, Q3
• NYC Borough Shape file

Methods/Results/Future Direction

Initially I started with the entire dataset, but wasn’t really sure what to do with it.

Using the shape file for congressional districts in cartodb’s dataset library, I thought I could show the number of institutions in each district and then another layer with each type of institution. For some reason this wasn’t working and I decided it was too much information so I switched gears to focus just on the NYC area and the five boroughs. I was able to find the borough map shape file in the CartoDB dataset library and easily add to my account. It was initially all orange. I played around with different color combinations but eventually settled on using one of the preset color combos in the library, which is a range of mostly blues with some yellow. After our class on Visual Perception, Color, and Narrative Design, I knew that blue was not one of the most highly visible colors. It made sense to use as the underlying color for the boroughs to me. I adjusted the transparency so that it was not opaque. I also switched to the base map without labels (the one for Queens was not showing). The labels are from the Borough shape file and adjusted for better visibility. I played around with the font and the outline and tried to make it clear what region was what. Although most New Yorkers are probably familiar and know each of the areas, I wanted to make it usable and clear where each borough was.

I then created a layer with each type of institution. The original dataset has nine categories. I merged them so there were six categories. Ideally I would like to have these as filters. This would allow a user to have a sense of how many of each type of institution there are in the five boroughs and if there are more of one type than another in certain areas.

Next I created a third layer related to the revenue. I edited the excel file so that any institution that I did not have revenue information, I removed. This pretty much halved my row count form approx. 400 to approx. 200. I created a revenue range and categorized each institution. I used the wizard and the legend template to set the colors, but had some difficulty getting these set right. In the end, I ended up going with blue and purple color ranges (what I didn’t like about the 2015 Art and Politics map above, but it worked okay here). I wanted this layer to be visually distinct from the other layer of institution type so I used a visually distinct color palette. Ideally a viewer would only look at two layers at a time (borough map and institution type or borough map and revenue). Viewing all three would be too much and would obscure things. Again for the future, I think I would create a filter for each of the categories. There is only one institution in the revenue filter with “1 Billion +” and it is obscured by other dots. In the future I would like the search function to work for finding legend titles, but maybe that only works with filters, I will need to explore more. I find the zoom scroll feature helpful for zooming and liked that the Alternative Fuel map used it too.

Ultimately I think cartodb works better with shaded shape maps than points, like this and this. At least that’s my thought at the moment unless you are focused in on a small area. Maybe it is best to combine the shaded areas and points- depending on the data. I think my viz would be more successful if I had another layer with smaller areas defined (maybe zip codes or census areas) and the filters. Another future consideration would be using the map as a narrative, I liked this example and could see using that for a different type of project all together.