Outliers in Smartphone Sensor Data Reveal Outliers in Daily Happiness

Enabling smartphones to understand our emotional well-being provides the potential create personalised applications and highly responsive interfaces. However, this is by no means a trivial task - subjectivity in reporting emotions impacts reliability of ground-truth information whereas smartphones, unlike specialised wearables, have limited sensing capabilities. In paper, we propose new approach that advances state prediction extracting outlier-based features mitigating capturing information. We utilised distinctive challenging use case happiness detection demonstrated performance improvements up 13% AUC 27% F-score compared traditional modelling approaches. The results indicate extreme values (i.e. outliers) sensor readings mirror reported levels. Furthermore, showed more robust replicating model completely experimental settings.