It is occasionally challenging to secure the cyber-physical systems (CPS) that underpin the Internet of Things (IoT) since security measures taken into account for general information and technology operations (IT/OT) might not function well in a CPS environment. In this article, we suggested an automatic cyberattack detection and classification system built around machine learning. It uses several ML models to achieve IoT cybersecurity applications. Learning based Cyber Attack Detection and Classification (LbCADC) is the approach we suggested. An ML pipeline and an IoT integrated use case dataset are the algorithm's inputs. It trains classifiers using supervised learning, and then uses the generated models to automatically detect and categorize cyberattacks. An empirical study evaluates our methods. The results of the experiments showed that the ML models could identify and categorize cyberattacks. The best accuracy of 98.36% was demonstrated by Multilayer Perceptron (MLP) in binary classification. SVM performed best for multi-class classification, achieving 97.59% accuracy.