Upon attempting to create the 5 different Word clouds that would be used to form the large map of Pittsburgh however, I ran into quite a bit of hiccups.
Limitations when sorting the data:
1. Postcodes jump boundaries!.... and districts! In some cases one zip code is on both sides of the rivers. This proved to be very difficult when trying to sort the data to demonstrate the different characteristics of regions, let alone my original plan to do it based on neighborhood boundaries.
2. Some zip codes are almost fully outside of the PGH boundaries, but may still apply for tweets within the boundary. I made the choice to keep any zipcode data that fell within a region boundary.
3. Because of the overlap, some zipcode data was doubled. For instance, 15207 falls both in Hazelwood (which is in the East) and also in New Homestead (in the south)
Limitations with Wordaizer App:
1. There wasn't enough data to be read by the Wordaizer map for all regions.
2. Because the amount of words were too repetitive (because the data was scraped from just one week of tweets), the map was very open and loose.
Ultimately I had to combine all region data to achieve a denser map that when combined could remotely look like Pittsburgh. My final results were created using one text file with all occurrences of words pasted to a Pittsburgh master-file. I then created a bitmap for Wordaizer of the PGH boundaries.
Is this a good/useful/informative piece of content to include in the project? Have your say!
You must login before you can post a comment. .