Background: Many users rely on ratings of a restaurant to choose where to dine. Conversely, while not a measure of the sustainability of the restaurant as a business, a good rating suggests that there is demand for that type of food in that area. Relating restaurant ratings to geospatial factors like distance from the city center, universities, as well as demographic factors would help reveal why a certain type of food or restaurant is popular in a particular area. These relationships may also provide a means for "normalizing" restaurant scores across cities, making it easier for customers to pick restaurants.

Data collection: Google APIs exist, but are not free. Scraping is an option, with multiple scraping tools available specifically for Google Maps. The key problem is the sheer number of restaurants even in a small town. We would likely be constrained to a small geographic location, or compare multiple city centers scattered across the US. Other data like demographics is also likely scrapeable, at the very least from sources like Wikipedia.

Analysis: Selecting features is likely tricky. What I propose to add to the Google Maps data with ratings the following features: nearest school distance, nearest restaurant of the same type of cuisine, median household income in the area, percentage of population belonging to different