Monday, April 29, 2013

Your smartphone logs- data for defining your personality traits


I have in one of my earlier posts covered the ability of leveraging a user's Facebook Likes to predict the user's personality type. Today's post covers a similar ability to accurately predict the smartphone's user personality from the the logs available on the phone.
(Note: Logs are available to any app installed on a smartphone.)

 In a paper titled 'Whos who with Big-Five: Analyzing and Classifying Personality Traits with Smartphones'  the authors present compelling evidence that data from smartphones can very accurately bucket the user for any trait as defined in the big five personality traits system. The big five personality traits are defined here for easy reference.

  • Extroversion - Active, Assertive, Energetic, Enthusiastic, Outgoing, Talkative 
  • Agreeableness - Appreciative, Forgiving, Generous, Kind, Sympathetic 
  • Conscientiousness - Organized, Reliable, Responsible, Thorough 
  • Neurotic-ism - Anxious, Self-pitying, Tense, Touchy, Unstable, Worrying 
  • Openness to Experience - Artistic, Curious, Imaginative, Insightful, Original, Wide Interests 



The authors were able to use pretty standard analytical techniques on data collected from a rather sample small size of 117 volunteers in Switzerland using a certain model of smartphone. The study would have been more useful if the sample size was higher.

The authors looked at the following data points collected by a monitoring app running on the volunteer's smartphone
  • Call Logs
  • SMS Logs
  • Bluetooth scans via BT Logs
  • Calling profiles (Ringtone, etc)
  • Application usage via App Logs (for Mail, Youtube, Music) 
Based on the above data sets, the authors were able to determine the High/Low rating for the big five traits that matched the self assessment provided by the volunteers. Some of the more interesting findings from the paper that highlight gender differences:

"It was found that extraverts, who are characterized by talkativeness and outgoing nature, were more likely to receive calls and also spend more time on them. Features pertaining to outgoing calls were found not to be predictive of the Big-Five traits. Agreeableness among females was associated with an increase in the number of incoming calls. Agreeable males were found to communicate with more number of unique contacts through voice calls. On the other hand, conscientiousness was associated with higher usage of the Mail app, that could be used in a professional context, and with lower usage of the Youtube application, which is likely to be used for entertainment purposes. Conscientious users were also likely to contact lesser number of unique people through voice calls. This conforms with their characterization in the literature as responsible and organized individuals. Interestingly, emotional stability was liked to higher incoming SMS. And high openness was associated with increased usage of Video/Audio/Music apps in females and also with the usage of non-standard calling pro files such as Beep..." 

As smart devices invade more into our lives easier they make for machine learning systems to predict who we are.  This knowledge is a goldmine for a marketer or a recruiter.