Adi Wibowo

The rapid development of Internet of Things (IoT) technology in smarthome systems has significantly increased the demand for convenience, efficiency, and security. However, this growth also poses serious cybersecurity challenges, especially with attacks targeting IoT devices, which are often inadequately secured. Threats such as Denial of Service (DoS), Distributed Denial of Service (DDoS), and Man-in-the-Middle (MitM) attacks have become real risks that compromise the integrity of smarthome systems. This research aims to develop an adaptive and efficient cyber attack detection system based on machine learning. The main focus is the implementation of the XGBoost algorithm to detect various types of attacks on smarthomes, with comparisons made to other methods such as Support Vector Machine (SVM) and Random Forest (RF). The COMNETS dataset is used as the experimental basis, as it contains cyber attack types relevant to IoT environments. The research process includes data preprocessing, feature selection, model training, and performance evaluation using accuracy, precision, recall, and F1-score metrics. Feature selection methods include Information Gain, Gain Ratio, and Chi-Square. Initial results show that XGBoost consistently outperforms other algorithms, achieving 94.8% accuracy, 95.2% precision, 94.5% recall, and 94.7% F1-score. This study is expected to contribute significantly to the development of Machine Learning -based cybersecurity systems capable of identifying and responding to cyber threats in smarthome environments, especially in the Indonesian context. Keywords: Smarthome, IoT Security, Machine Learning, XGBoost, Cyber Attack Detection