twimoji

a mostly-for-fun exercise in analyzing location-enabled tweets with emoji in Philadelphia

Data

Collection

This project uses the simplest method to aggregate tweets based on simple parameters:

• This IFTTT job to collect tweets with location attributes in a given radius around Philadelphia:

• Saves to a Google Spreadsheet, when full begins new sheet

Cleanup & Reshape

• Tokenize words from tweet text for possible later NLP
• Extract latitude & longitude from link to map field string

Analysis

• Emoji lookup table
• Map unicode and bytes from table
• Calculate word frequency
• Only count 1st occurence of an emoji per tweet

Visualization

Geocoding

• CartoDB

Location Filtering

• Uses city boundaries file (link)
• Add geodata for additional demographic analysis:
  • Census tracts or Block segments
  • Zip codes?

Time-based animation

• Torque.js