A recent study by the MIT Senseable City Lab has used cellphone data to map both social and physical segregation within Singapore. To start, they used residential sale prices as a proxy for socioeconomic status. They then used call and text records (presumably it was all anonymous) from 1.8 million cellphone users in Singapore (2011) to map who interacted with who. Pictured above is one of those mappings.
What they discovered was evidence of a "rich club effect." In other words, the richer the person the less likely they were to interact with people outside of their socioeconomic band. The study calls this their communication segregation index.
A similar phenomenon was noted as people moved around Singapore. (This is the study's physical segregation index.) People tend to spend time in spaces alongside people with similar socioeconomic attributes. However, they did notice that this tends to wane during the day as people move around the city -- presumably for work and other such things.
I think it would be interesting to get a bit more granular about the findings in order to try and see, among other things, if certain public spaces are more successful than others at encouraging a broader socioeconomic mix. And it's probably only a matter of time before we start using tools like this to plan our cities. For more on the study, click here.
Image: MIT Senseable City Lab
A recent study and research paper by the MIT Senseable City Lab -- called, Tasty Data -- has discovered that restaurant data alone can be used to accurately predict location-based factors such as daytime population, nighttime population, number of businesses, and overall consumer spending within a specific geography.
They started by pulling restaurant data from Dianping (Chinese equivalent of Yelp) for 9 Chinese cities: Baoding, Beijing, Chengdu, Hengyang, Kunming, Shenyang, Shenzen, Yueyang, and Zhengzhou. They then paired their Dianping data with other available data (such as aggregated mobile phone data) and used machine learning to search for any correlations.
Below is a diagram of "nighttime population" in Beijing. They are using a 3 km2 grid.
The latest project out of MIT's Senseable City Lab examines the "sensing power of taxis" in various cities around the world. Looking at traffic data, they determined how many circulating taxis you would need to equip with sensors if you wanted to capture comprehensive street data across a particular city. This might be useful if you wanted to measure things like air quality, weather, traffic patterns, road quality, and so on.
What they found is that the sensing power of taxis starts out unexpectedly high. It would only take 10 taxis to cover 1/3 of Manhattan's streets in a single day. However, because taxis tend to have convergent routes, they also discovered rapid diminishing returns. It would take 30 taxis (or 0.3% of all taxi trips) to cover half of Manhattan in a day, and over 1,000 taxis to cover 85% of it. A similar phenomenon was observed in the other cities that they studied: Singapore, Chicago, San Francisco, Vienna, and Shanghai.
However, if you look at the percentage of trips needed to scan half of the streets in a city, Manhattan has the lowest rate at 0.3%. Vienna is the highest at 9%. But I'm not sure if this is a function of the utilization rate of their taxis or if it has something to do with urban form. Singapore has a similarly low rate (0.44%), but its street grid looks nothing like that of New York's.
Here's a short video explaining the project:
https://youtu.be/Vs3q3jQaM9Q
A recent study by the MIT Senseable City Lab has used cellphone data to map both social and physical segregation within Singapore. To start, they used residential sale prices as a proxy for socioeconomic status. They then used call and text records (presumably it was all anonymous) from 1.8 million cellphone users in Singapore (2011) to map who interacted with who. Pictured above is one of those mappings.
What they discovered was evidence of a "rich club effect." In other words, the richer the person the less likely they were to interact with people outside of their socioeconomic band. The study calls this their communication segregation index.
A similar phenomenon was noted as people moved around Singapore. (This is the study's physical segregation index.) People tend to spend time in spaces alongside people with similar socioeconomic attributes. However, they did notice that this tends to wane during the day as people move around the city -- presumably for work and other such things.
I think it would be interesting to get a bit more granular about the findings in order to try and see, among other things, if certain public spaces are more successful than others at encouraging a broader socioeconomic mix. And it's probably only a matter of time before we start using tools like this to plan our cities. For more on the study, click here.
Image: MIT Senseable City Lab
A recent study and research paper by the MIT Senseable City Lab -- called, Tasty Data -- has discovered that restaurant data alone can be used to accurately predict location-based factors such as daytime population, nighttime population, number of businesses, and overall consumer spending within a specific geography.
They started by pulling restaurant data from Dianping (Chinese equivalent of Yelp) for 9 Chinese cities: Baoding, Beijing, Chengdu, Hengyang, Kunming, Shenyang, Shenzen, Yueyang, and Zhengzhou. They then paired their Dianping data with other available data (such as aggregated mobile phone data) and used machine learning to search for any correlations.
Below is a diagram of "nighttime population" in Beijing. They are using a 3 km2 grid.
The latest project out of MIT's Senseable City Lab examines the "sensing power of taxis" in various cities around the world. Looking at traffic data, they determined how many circulating taxis you would need to equip with sensors if you wanted to capture comprehensive street data across a particular city. This might be useful if you wanted to measure things like air quality, weather, traffic patterns, road quality, and so on.
What they found is that the sensing power of taxis starts out unexpectedly high. It would only take 10 taxis to cover 1/3 of Manhattan's streets in a single day. However, because taxis tend to have convergent routes, they also discovered rapid diminishing returns. It would take 30 taxis (or 0.3% of all taxi trips) to cover half of Manhattan in a day, and over 1,000 taxis to cover 85% of it. A similar phenomenon was observed in the other cities that they studied: Singapore, Chicago, San Francisco, Vienna, and Shanghai.
However, if you look at the percentage of trips needed to scan half of the streets in a city, Manhattan has the lowest rate at 0.3%. Vienna is the highest at 9%. But I'm not sure if this is a function of the utilization rate of their taxis or if it has something to do with urban form. Singapore has a similarly low rate (0.44%), but its street grid looks nothing like that of New York's.
Here's a short video explaining the project:
https://youtu.be/Vs3q3jQaM9Q
If you're a regular reader of this blog, you'll know that I like these kinds of studies. By 2020, it is estimated that 1.7MB of data will be created every second by every person on earth. The numbers are staggering. And yet, "official" data sources, such as census data, remain slow and fairly limited. Studies like this one continue to show us what's next.
Image: MIT Senseable City Lab
If you're a regular reader of this blog, you'll know that I like these kinds of studies. By 2020, it is estimated that 1.7MB of data will be created every second by every person on earth. The numbers are staggering. And yet, "official" data sources, such as census data, remain slow and fairly limited. Studies like this one continue to show us what's next.
Image: MIT Senseable City Lab
Share Dialog
Share Dialog
Share Dialog
Share Dialog
Share Dialog
Share Dialog
