Explicative human activity recognition using adaptive association rule-based classification

摘要: Computational social sensing is enabled by the Internet of Things at large scale. Using sensors, e. g., implemented in mobile and wearable devices, human behavior activities can then be investigated, e.g., using according models patterns. However, obtained are often not explicative, i. e., interpretable, transparent, explanation-aware, which makes assessment validation difficult for humans. This paper proposes a novel explicative classification approach featuring interpretable explainable models. For this purpose, we embed framework building rule-based classifiers class association rules. evaluation, apply two real-world datasets: One collected domain personalized health sensors (accelerometers), second one utilizing smartphone activity recognition. Our results indicate, that proposed outperforms baselines clearly, concerning both accuracy complexity resulting predictive