Outliers in Smartphone Sensor Data Reveal Outliers in Daily Happiness
summary
Smartphone sensor readings are converted into per-user outlier encoding (lower/regular/upper) and the daily happiness label is also encoded per-person, then standard classifiers are trained on these binary outlier targets. Evaluation uses phone usage logs and sensors from 221 diverse + 481 university participants (Android only), daily EMAs and one happiness question; the outlier features beat traditional aggregates, yielding up to 13% AUC and 27% F-score improvements and generalizing when training on the diverse cohort and testing on the university cohort.
AI-written summary · about
related talks
26m"Reality Mining for Compan
23mKeynote: Digital Psychometrics and it's Future Effects on Technology
17mKnow Thyself: Introspective Personal Data Mining
32mUnderstanding and Measuring User Emotions: The Missing Puzzle Piece
16mYour Mind, Quantified
10mIntelligence in the Age of Plethora
06mMachine Learning for Phone-Based Relationship Estimation: The Need to Consider Population ...
05mComparing Security and Privacy Attitudes Among U.S. Users of Different Smartphone and