MapBox has posted an interactive Twitter visualization that shows a heatmap of smartphone Tweets by region, broken down by smartphone brand. It shows the iPhone (red) dominating in affluent areas and Android (green) in poor regions. Blackberry (purple) use is predominantly outside of major cities and in Africa, Central America, Southeast Asia and the Middle East. Pictured above is a map of Washington, Philadelphia and New York showing a heavy bias toward Tweets from iPhones.
Mobile Devices is a map that reveals the information about phone brands that is stored when people use an official Twitter App and is hidden in the metadata attached to each tweet. Each brand of phone is a different color and can be independently toggled. The patterns of usage in each city often reflect economic stratification. For example iPhones, in red, are predominantly in wealthy sections of the city while Android phones, in green, have more coverage in poorer sections. On a global level, national trends reveal a complicated set of cultural preferences. (Tweets from web browsers and from other Twitter clients don't appear on this map)
The screenshot above visualizes tweets from Locals and Tourists. It analyzes behavior over time to highlight areas of cities popular with locals and places that are usually documented by tourists.
This MapBox blog post talks about the how it visualized 3 billion Tweets:
This is a look at 3 billion tweets - every geotagged tweet since September 2011, mapped, showing facets of Twitter's ecosystem and userbase in incredible new detail, revealing demographic, cultural, and social patterns down to city level detail, across the entire world. We were brought in by the data team at Gnip, who have awesome APIs and raw access to the Twitter firehose, and together Tom and data artist Eric Fischer used our open source tools to visualize the data and build interfaces that let you explore the stories of space, language, and access to technology.
This is big data, and there's a significant level of geographic overlap between tweets, so Eric wrote an open-source tool that de-duplicated 2.7 billion overlapping datapoints, leaving 280 million unique locations.
Search the interactive Tweet map for your city. What conclusions can you draw?