Twitter Evacuation Patterns

From Micro -> Macro. For each particular user, we are getting a good picture of their individual actions, but we can’t generalize this data well to a macro picture to understand what the larger population did. In sum, Our current visualization efforts aren’t telling the story in a macro-digestible fashion.

Ergo, how do we find more prominent patterns?

1. If looking at the graphics isn’t immediately obvious, then we need new models? Markov Chain model? One state of preparation to another?

2. We are taking a purely data driven approach and it’s currently proving inconclusive, perhaps we need a larger RQ to investigate directly?

Current Problems

1. Timeline Visualization is unclear as-is.
   • What is next? Even if we can make pattern from noise, what are we looking for?

Current Successes

1. Frequency Graphs tell a better story
2. Coding Scheme is collapsed

Todo by Dec. 18

1. Update Frequency Graphs to show new frequencies
2. Line up all frequency graphs to see this story
3. Redo the timeline with better icons & resolution (in visjs)

Next Steps

1. Identify new people + contextual streams from new study areas to code, focused NYC study areas:
   • Manhattan Evacuation Zones
   • Rockaway Beach
   • Staten Island