An intelligent approach for human activity analysis using data mining techniques

An intelligent information system for sensor-based human activity analysis using data mining techniques presents a comprehensive study on developing a system that employs data mining to analyze human activities based on sensor data. With advancements in wearable technologies and embedded sensor systems, such as smartphones, smartwatches, and various environmental and object-attached sensors, it is now possible to automatically and continuously recognize and track human activities through sensor data collection. These techniques are generally known as sensor-based human activity analysis and can be applied across multiple fields, including healthcare, entertainment, and artificial intelligence system design. The core approach involves abstracting sensor data into higher-level activity recognition through various data processing and mining methods. In this study, eight classifiers are applied to the HARSense dataset, including naïve Bayes (NB), decision tree, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), neural network, XGBoost, random forest (RF), and extra trees classifier (ETC). The models are evaluated on the HARSense dataset, with the extra trees classifier achieving the highest accuracy of 97.12%.