Multi-Sensor IoT Smart Home Anomaly Detection Using Random Forest Algorithm

Abstract

The increasing adoption of Internet of Things (IoT) technology in smart home environments generates large volumes of dynamic and heterogeneous multi-sensor data, creating challenges in accurately detecting anomalous conditions. This study aims to implement and evaluate the Random Forest algorithm for anomaly detection in IoT-based smart home systems using multi-sensor data. The dataset consists of temperature, humidity, light, motion, carbon monoxide (CO), liquefied petroleum gas (LPG), smoke, door/window status, and energy consumption collected from three IoT devices. Data preprocessing included cleaning, labeling, and Min-Max Scaling normalization, followed by an 80:20 training-testing split. Model performance was evaluated using a confusion matrix, accuracy, precision, recall, and F1-score. The experimental results achieved an accuracy of 96.75%, precision of 94.65%, recall of 93.82%, and F1-score of 94.23%, demonstrating that Random Forest is effective for identifying anomalous conditions in smart home IoT environments.