Zoning the city: Infering land use from mobile phone activity

In order to better allocate scarce resources for long term development projects, more information is needed on how residents move through and use a city. Traditional data sources such as zoning regulations and travel surveys are often idiosyncratic, expensive, and infrequently updated. Mobile phones offer a far richer data source to measure the urban environment. Using anonymized location data for millions of phone events within a city, we use machine learning techniques to classify zoning across the city with high rates of accuracy.