metAtlas: Metropolitan Atlas from social media

Work done as Project Assistant, at the MIT Media Lab / Macro Connections Group (guided by prof. César Hidalgo). Work included the download and pre-processing of twitter geo-located data for 50 Metropolitan Areas around the world (7-8 million tweets), the development of a socio-spatial Clustering algorithm for definition of urban areas (based on Spectral Analysis of networks), and the Assistance in the writing of proposals for securing 30K+ US$ in funding for the project (MIT MISTI-Chile Grant).

Twitter Density Maps

Examples of Twetting Density in a 1 KM X 1 KM pixel lattice, for various cities. Maps not to scale.

Back to Twitter Maps

new york-new jersey-philadelphia, N° Tweets: 2MM

tokio-yokohama, N° Tweets: 2MM

minneapolis-st paul

lower countries, N° Tweets: 570K

london, N° Tweets: 500K

Twitter Rythms

Hourly Tweeting frequency, during one week for selected cities and/or regions.

Back to Twitter Maps

lower countries, N° Tweets: 570K

toronto-buffalo, N° Tweets: 300K

minneapolis-st paul

buenos aires-la plata, N° Tweets: 380K

chicago, N° Tweets: 500K

Clustering Experiments

Applying PCA to socio-spatial networks constructed from the data, possible urban agglomerations areas are explored. Maps are not to scale.

Back to Twitter Maps

0-tweeting density by pixel

1-first eigenctor score

2-liverpool area

3-manchester-area

4-liverpool (blue) and manchester (red)

Other Clustering Experiments

Using the same process outlined above (without PCA), possible clusters are explored for selected poly-centric urban agglomerations. Maps not to scale.

Back to Twitter Maps