Low Hanging Data Ripe for Predicting

Alex Pentland of MIT Laboratory is showing us a map of San Francisco that has different colored dots representing people categorized by behavior. The “green dot people” go to place “x” and then typically frequent place “y” and “z” whereas the “blue dot people” and “red dot people” have different behavior patterns. He imagines a world where we can use accurate real-time data to provide goods and services more efficiently and effectively to the American people.
For example, we could have a dynamic transportation system designed to meet the needs of people based on their movement patterns. What if a bus company could provide a service moved a group from place x to y to z, based on their needs, rather than a random bus schedule? This group of people would be spared the use of 20 different cabs when they’re all going to the same place. They just don’t know it yet.
The same could be true for neighborhoods. Neighborhoods have a different behavioral population at noon on Saturday than they do at 2pm on Tuesday. What if we could collect data to understand the behavior patterns of these groups in real time? Imagine the implications for public safety. We would be able to predict crime.
Now imagine the implications of behavioral data collection for health care. The streets of San Francisco are home to many populations. Although the people live in close proximity and have some similar demographics, they have very different needs and behaviors. For example, people identified by green dots on the map may be most likely to get swine flue. If we have a swine flu outbreak, health care workers would want to get to the “green dot” people first and implement prevention and recovery measures in places they frequent.

Leave a Comment

One Comment

Leave a Reply