Food Desert Server

This repository contains a server and javascript web client that attempt to detect and display areas that are in a food desert. The current implementation achieves this by querying the Google Places API for the locations of grocery stores then storing this data in a Spatialite database. Road network data is obtained from OpenStreetMap.

For the purpose of this project, a point is in a food desert if there are no grocery stores contained in a 1 mile buffer around the point. This approach is limited due to the quality of data obtained from the Places API and the inadequate size of a 1 mile buffer outside of densely populated urban areas. Improving the data source and increasing the size of the buffer outside of urban areas would greatly improve the usefulness of this project.

Setup Project

• Install sqlite 3.24
• Install spatialite 4.3.0
• Install JDK8
• git clone https://github.com/jackastner/FoodDesertServer.git
• Enter a google api key valid for use with the Google Places API into the file google_api_key in the root of this repository.
• ./gradlew test

If the Gradle build succeeds and all tests pass, the project has been successfully setup. The most common reason for failing tests is a missing or invalid API key.

Prepare Road Network Database

The server uses data from Open Street Map to compute a 1 mile walking distance buffer around grocery stores. Since this data can be extreamly large and areas of interest depend on the user, it is not included in the repository; however, it is available for free download from Open Street Map.

• Visit the OSM export website. Use the controls to select an area of interest then download the data from OSM or one of their mirrors.

• Process the data with the provided mknetwork script. This script runs the Spatialite tool for processing OSM data then performs post processing to transform the data into the EPGS used by the server.

    ./mknetwork network.db map.osm
  

Run Server

First, follow the the steps in Project Setup. You can then choose to run the server directly through Gradle or by building and executing a jar file.

Run with Gradle

./gradlew run This will start the server using the google api key entered during setup and using the default database files (data.db and network.db).

Build the Jar

• ./gradlew jar

• java -jar build/libs/FoodDesertServer.jar to see usage information.

    Usage: java -jar FoodDesertServer.jar database_file network_database_file [google_api_key]
        database_file: SqLite database file containing tables created by this server.
        network_database_file: SqLite database file containing tables created by spatialite_osm_net
        google_api_key: a valid key for the Google Places API. If omitted, this
            program will look for a Java properties file containing a key value pair:
            google_api_key=$YOUR_API_KEY
  

• java jar build/libs/FoodDesertServer.jar data.db network.db to run in the default configuration.

Use Server

Once the server is running, the web interface is available at http://localhost:4567/food_desert_map.html. The interface provides access to methods defined in FoodDesertQueryHandler and displays results on a map.

---

Unimplemented Features

• Obtain and display polygon(s) representing areas that are food deserts (Completed but subpoints aren't).
  • Generate these polygons using walking distance (rather than Euclidean distance) (in progress).
  • Account for location in these polygons. In Suburban and rural areas, distance to a grocery store can be much larger without being a food desert.
• Find alternative data sources. Google Places API will be paid only soon (It still works for some reason).
  • PG county data website provides grocery store locations that could be loaded into the database. Presumably other data sources like this exist.
  • I would like to find a solution that can get data for arbitrary locations like the Places API.
• Switch user interface to OSM.
• Document API.
• Implement front ends for other platforms (IOS/Android/mobile web).