Claremontrun is an open-source Github repository that holds numerous datasets compiled as a part of a project tracking Claremont’s X-Men publications. There are seven datasets that have been collected in regard to this project and they all track different aspects of his issues. Here are some examples of information they hold: character information, such as their actions; locations that appear in each issue; data about issue covers; bechdel test data. My project, of course, made use of the Locations dataset. The Location data originally only had four columns–“# Issue”, “Location”,  Context (e.g.”Present “,”Dream “, or “Past”), and Notes– formatted as an XLSX which contained 1,592 rows. The “Location” values are strings that either discretely describe location (e.g. Jean Grey’s Apartment, Greenwich Village, NYC), or vaguely give the location setting (or object) that the characters in (e.g. “Asteroid M, Space”, or “Plane flying over Canada”). This type of string data made it too difficult to parse in Excel or OpenRefine, so I converted it to a CSV and wrangled it using R/RStudio. I used ArcGIS Online to create my heat map and supplemental pie chart. 

Presentation

I decided to use a simpler website to share my information. I created an Instant App through ArcGIS and embedded that into my website. For some reason, it at first only shows the chart, but viewers can navigate to the map themselves, or click on the provided link to an external view. I have my map and text side by side, as I thought that it might be helpful to read the process while glancing at the map to see the direct results of my decisions. I’ve also included my data at the bottom of the page so that others can use it for themselves if they ever wanted. The original data can be found in Github.

Data Processes

There was lots to do when it came to cleaning up this data. For starters, I removed the “Notes” column as it contained no information relevant to my goals, and narrowed the scope of “Context” so that I was only focusing on “Present” locations. I thought there was a possibility that past or dream contexts could’ve referred to locations that were part of the X-Men series, but maybe not in the issues written by Claremont. Furthermore, I came up with a few criteria for which location data I could keep based on clarity, and whether it could actually be recognized and mapped in ArcGIS online. I used keywords and string detection to process the data in this way. Any location description that contained a word related to Space, Planet, Orbit, Spaceship, Aircraft, Plane, Dimension, or some sort of motion, I removed. Next, I wanted to standardize most of the locations so that a location such as “Stornoway, Scotland’s North-West Coast” would become “Stornoway, Scotland”. For this, I used spacyr, an R package with Named Entity Recognition (NER) which recognized and extracted locations names of cities, states, or countries. This step in my processing was where I ran into problems with OpenRefine, and decided to switch to a tool I was more familiar with. These choices left me with a much smaller, but cleaner dataset: 582 rows of mappable data. My final dataset had four columns: “Issue #”, “Location”, “Context”, and “Place” which stores all the standardized location data.