Location-Tracking Mobile App

Final Results

A Location-Tracking Mobile App

Advisor: Jeff Ondich
Times: Winter 3a

Location data is powerful. If you know an individual's location every 15 minutes over a period of several months, you can identify with high probability where the person lives and works and goes to school, where they go for fun, what kind of medical care they receive, etc. If you have a database of many people's locations, it's easy to search for friend groups, work groups, participants in political protests, weekly meetings, and other kinds of relationships (illicit or not). Even if the data is ostensibly "anonymized", location data is so particular to an individual that it's very easy to translate a detailed stream of locations into a single individual's identity.

Law enforcement and intelligence agencies love location data, and there is an extensive, largely unregulated marketplace making large datasets of location data available for sale.

Sometimes, sharing your phone's location with an app leads to great features for you. Mapping apps like Google Maps, Apple Maps, and Waze are very popular for good reason—they help you get from your current location to your destination, they help you route around traffic delays, etc. And fitness apps like Strava and Runkeeper can give you helpful guidance about your exercise. But because the data market is mostly unregulated, the creators of these apps can also sell your location data or put it to uses unrelated to the purposes of the apps that collected the data in the first place.

The problem with location data is that it can lead to extraordinarily intrusive, detailed inferences about the behavior of individuals and groups of people. Despite the fact that this has been widely studied and publicized, there's still a surprising amount of apathy around the subject of data collection and exploitation.

To help us understand more concretely how location data can be translated into inferences about a person, this project will focus on creating a location-tracking mobile app. This app will collect its user's location over time and provide the user with a variety of data visualization and analysis features intended to demonstrate the power of this kind of data.

What you will do

Decide whether to create an app for iOS or Android or both.
Study privacy policies and write one suitable for your app.
(Weeks 1-4) Use the built-in iOS/Android location services to create an app that collects location data at fixed intervals (say, 5 or 15 minutes), stores it in a local database on the phone, includes very simple data visualization (e.g., a daily map and a tabular list of GPS coordinates), and allows export of the data as, say, an emailed CSV file. At this stage, the app should also include an On/Off setting to allow the user to easily enable and disable data collection.
Incorporate existing location databases (e.g., from Google Maps or OpenStreetMap) to tie the user's GPS record to named locations like specific businesses, restaurants, street addresses, etc.
Conceive, design, and implement a suite of analytical tools for the data. For a simple example, your app could try to identify the location of the user's home. You could show a list of the user's top-ten locations (by name and coordinates), predict the user's likely location at a future date and time, etc. etc. There's a lot of room for creativity in this space.
Design and execute a few simple tests for your analytics (e.g., how effective is the app at detecting people's homes?). Because we won't have the time to collect medium-to-large datasets for this project, we won't be shooting for formal statistical studies during this comps period. But we want to do some simple some plausibility tests to make sure our analysis features aren't obviously dumb, which could help set up the potential for more formal studies in the future.
(Optional) Enable two users to share their data with each other and provide two-person analytical features like "how much time do they spend together?" and "how far apart do they live?"

(Also, if you like combining entertainment with existential dread, you should probably watch the Data Brokers episode of John Oliver's Last Week Tonight.)