The Impact of Domain Shift on Predicting Perceived Sleep Quality from Wearables.

Machine learning models for personal informatics systems are typically trained offline on , resulting in These models may suffer performance degradation in real-world settings due to , i.e., differences in data distributions across users and contexts. Domain adaptation techniques can address this by, personalizing models with user-specific data. on of both population and personalized models sleep quality recognition. . Our analysis shows domain shift accuracy of population models by up to 18.54 percentage points, when on . Personalized models, , show robust performance across datasets. However, , limiting their . To the limitations , we propose a novel unsupervised domain adaptation approach: the cluster-based population model (CBPM). CBPM achieves accuracy improvements of up to 13.45 percentage points without requiring user-specific records or .