The Internet of Things (IoT) comprises a network of interconnected nodes constantly communicating, exchanging, and transferring data over various network protocols. Studies have shown that these protocols pose a severe threat (Cyber-attacks) to the security of data transmitted due to their ease of exploitation. In this research, we aim to contribute to the literature by improving the Intrusion Detection System (IDS) detection efficiency. In order to improve the efficiency of the IDS, a binary classification of normal and abnormal IoT traffic is constructed to enhance the IDS performance. Our method employs various supervised ML algorithms and ensemble classifiers. The proposed model was trained on TON-IoT network traffic datasets. Four of the trained ML-supervised models have achieved the highest accurate outcomes; Random Forest, Decision Tree, Logistic Regression, and K-Nearest Neighbor. These four classifiers are fed to two ensemble approaches: voting and stacking. The ensemble approaches were evaluated using the evaluation metrics and compared for their efficacy on this classification problem. The accuracy of the ensemble classifiers was higher than that of the individual models. This improvement can be attributed to ensemble learning strategies that leverage diverse learning mechanisms with varying capabilities. By combining these strategies, we were able to enhance the reliability of our predictions while reducing the occurrence of classification errors. The experimental results show that the framework can improve the efficiency of the Intrusion Detection System, achieving an accuracy rate of 0.9863.