The training data used to develop the classification algorithms and train them The training data. This was collected using the sensor logger application.

It’s a basic table with just two columns - log_headers and log_data. The headers field contains selected raw HTTP headers which the app uses to communicate metadata like the selected activity, app version, IMEI number etc. In the export the IMEI number is replaced with a unique user ID to avoid data protection issues. The log_data field contains one line for each sample recorded, in the same format to the other file. The newer versions of the app record the orientation sensor as well, so these will have 9 data points total instead of 6.