This morning I started reading a research paper by MIT researchers César A. Hidalgo and Lisa E. Castañer called Do we need another coffee house? The amenity space and the evolution of neighborhoods (2015, PDF Link). The concept of an Amenity Space is a potentially powerful concept as it can be used to recommend what amenities might be missing, or not, in a neighborhood. The concept leads to a potential discovery technique for entrepreneurs and investors to find new businesses that might be viable in a given neighborhood.

One of the authors, Cesar Hidalgo, recently wrote a book called Why Information Grows: The Evolution of Order, from Atoms to Economies (2015) that looked like it might be interesting.

I decided to read the article to get some exposure to Cesar's thinking and because I was exploring the Atlas for Economic Complexity and realized he was also a driving force behind that impressive visualization project.

In this article, the authors used Google Maps to gather data on the amenities available in various neighborhoods. They computed the correlations between the amenities (how strongly the presence of one amenity predicted another) and represented the strength of these correlations by the thickness of the links between the amenity nodes. You can read more about how the visualization was constructed at the bottom of the diagram.

The paper proposes an algorithm that generates recommendations for new amenities for a neighborhood. I recommend you read the paper if you want these details.

The concept of an amenity space might be useful not just for making recommendations for new neighborhood amenities but for explaining why certain destinations are good tourist destinations, why real estate prices are higher in certain neighborhoods, what cities might do to improve their neighborhoods, etc....

Another point the authors make in the paper is that neighborhoods often have a pattern of specialization (e.g., tourism amenities) which might suggest a particular type of amenity that would be a good fit. An example from my own experience is the relationship between world-class golf courses and the need for nearby airport facilities for private jets. One tourist amenity (world-class golf courses) drives the need for a complementary amenity (airport for private jets to land). High-end golfers apparently also like high-quality whiskey and a nearby single-malt whiskey distillery with lounge and lodgings benefited significantly from the arrival of the golf course to the neighborhood. The golf courses are driving a new evolution of amenities in this rural community. While the concept of amenity space was developed based upon business co-location data in larger cities, it might be useful for thinking about expected co-location patterns in rural areas as well.

It also appears that the recommender system might take the form of a lens model as linear regression techniques were used to make the recommendations.