EmoPhone: A Multi-Wave Dataset for In-the-Wild Mobile and Wearable Affect Sensing

This repository releases the raw data for D1, D2, and D3, three consecutive annual datasets collected as part of an in-the-wild affective sensing study from 2020 to 2022. Each dataset combines passive sensing from participants’ own Android smartphones and Fitbit wearables with dense in-situ affective state labels collected via the Experience Sampling Method (ESM). In total, the repository contains 53,139 ESM responses from 297 participants after quality control (328 participants recruited overall), captured in naturalistic daily-life settings. The dataset is intended to support research on moment-level affect prediction (e.g., valence, arousal, stress, and disturbance), personalized and cross-person generalization, cross-dataset transfer learning, and longitudinal analyses of behavioral and affective variability in mobile affective computing.