Mood Detection and Prediction Based on User Daily Activities

Studies show that mood states influence our daily life quality and activities, and this is not the only way around. Mood also changes because of how we spend our days. In this paper, we use data on users' daily lives (known as lifelog) to both detect and predict their mood. The states of mood in this paper are based on Thayer's two-dimensional model of mood. This is the first research to analyze in depth the physical data collected in lifelog and its link to determinants and effects of mood including biometrics, physical activities, sleep quality, diet and user's environment. Our study shows that such link exists and is significant.