Table 1.
Behavioral features extracted by QuantifyMyPerson.
| Feature | Description | |
| Calls | ||
|
|
mean_incoming | Average duration of incoming calls (seconds) |
|
|
mean_outgoing | Average duration of outgoing calls (seconds) |
|
|
tot_call_length | Total call duration (seconds) |
|
|
outgoing_call | Number of calls made |
| Brightness | ||
|
|
mean_time_usage | Average duration of a session of use (from screen switch on to screen switch off) |
|
|
number_switch_on | Number of times screen is switched on |
|
|
b_n (24) | Seconds of phone’s use from hour n–1 to n with hourly granularity over the whole 24 hours |
| Apps | ||
|
|
tot_kb_social | Kilobytes consumed in social app (Facebook, Instagram, Twitter, LinkedIn) |
|
|
tot_kb_communication | Kilobytes consumed in communication app (WhatsApp, Messenger, Telegram, Skype, Hangouts) |
|
|
tot_kb_navigation | Kilobytes consumed in navigation app (Chrome, Firefox, proprietary browser, Google, YouTube, Tripadvisor) |
|
|
tot_kb | Total kilobytes consumed in a day |
| GPS | ||
|
|
number_of_clusters | Number of places visited |
|
|
time_outside | Percentage of time spent outside the home |
|
|
location_variance | Variability in a participant’s location calculated as location_variance = (σ2long + σ2lat), where σ2long and σ2lat represent the variance of the longitude and latitude, respectively, of the GPS location coordinates |
|
|
Entropy | Measure of how uniformly a participant spends time at different locations. Let pi denote the percentage of time that a participant spends in location cluster i. The entropy of the participant is calculated as entropy = −(pi*log(pi)) |
|
|
visited_clusters | Latitude and longitude coordinates of the visited places according to the distance from home |
| Activity | ||
|
|
m_amp_n (24) | Average of the acceleration signal amplitude from hour n–1 to n with hourly granularity over the whole 24 hours |
|
|
s_a_n (24) | Seconds of high activity from hour n–1 to n with hourly granularity over the whole 24 hours |
|
|
s_r_n (24) | Seconds of low activity from hour n–1 to n with hourly granularity over the whole 24 hours |
|
|
percentage_activity | High activity/(high activity + low activity) |