Mobility Data Commons

This is a repository for publicly available anonymized and pseudo-anonymized mobility data from the ohmage Mobility Android app (source, app). The Mobility app collects a data point every minute or every five minutes, depending on user preference.

The data is available as CSV exports. There is a pseudo-anonymized example here and a anonymized example here.

The data can be found in the data directory of this repo. Each directory represents one user's data. The data has been cleaned and is valid CSV and JSON, but irregularities may exist. Each directory contains gzipped tar archives that contain chunks of data for one to six month periods. Very gappy time periods have been removed. Each archive is roughly 50 MB or less.

If you have questions, feel free to put in an issue. Otherwise, clone this repo and analyze away!

Element Dictionary

id	A unique integer identifier for this object given its presence in a list of other mobility data points.
uuid	As above, but it is a UUID.
time	The standard Unix UTC milliseconds from January 1, 1970 up to the time this point was created.
time_offset	The number of milliseconds to convert time to the timezone where the point was collected.
timezone	The human-readable "zoneinfo" timezone. May be null.
location_timestamp	The timestamp at which the location was determined, which may be different than the time the point was created. May be null.
latitude and longitude	The location of the user at the time of location_timestamp. May be null.
location_accuracy	This value may be null, but if it isn't, the Android developer docs state the following. <p>Get the estimated accuracy of this location, in meters.</p> <p>We define accuracy as the radius of 68% confidence. In other words, if you draw a circle centered at this location's latitude and longitude, and with a radius equal to the accuracy, then there is a 68% probability that the true location is inside the circle.</p> <p>In statistical terms, it is assumed that location errors are random with a normal distribution, so the 68% confidence circle represents one standard deviation. Note that in practice, location errors do not always follow such a simple distribution.</p> <p>This accuracy estimation is only concerned with horizontal accuracy, and does not indicate the accuracy of bearing, velocity or altitude if those are included in this Location.</p> <p>If this location does not have an accuracy, then 0.0 is returned.</p></td></tr>
location_provider	May be null. The system entity that supplied the location. The possible values are WifiGPSLocation:GPS - the GPS receiver. WiFiGPSLocation:Cached - a cached GPS lock from the GPS receiver. WiFiGPSLocation:Fake - a fake provider for testing. speed will always be 0.0 in this case. WiFiGPSLocation:Network - the Wi-Fi Network scan provided the location. WiFiGPSLocation:Network - the Wi-Fi Network scan provided the location, but the value returned is from a cache of Wi-Fi scans and possibly incorrect. You can check out the source code for WiFiGPSLocation here. Note that other values for location_provider are possible for this field, but the list above contains the values that are officially supported.
wifi_data	A list of Wi-Fi access points and the time at which the access points were collected. A typical list size is between 5 and 10 points though it can be larger.
scan	A list of ssid-strength pairs. Strength is a negative integer that represents stronger signal strength as it approaches zero.
speed	The speed (not velocity) of the mobile device compared to the previous GPS reading. Can be zero if obtaining a GPS lock was unsuccessful.
accel_data	A list of accelerometer readings. The list is typically between 30 and 50 triaxial points.
mode	The classified ambulatory mode based on our classification algorithm applied to this point. The value will be one of still, walk, run, drive, or error. The error mode occurs if the classifier cannot determine the mode based on the data for the point.