The intrusion is expanding along with the quick growth of internet users and Internet of Things (IoT) devices. Most devices today are internet-connected and use data transmission to communicate. In that situation, intrusion detection is crucial to protecting the internet-connected devices from attackers. To increase the accuracy of intrusion detection, this study examines attacks retrospectively. Four well-known machine learning methods are taken into consideration for identification in this study. To identify DDoS (types: HTTP, TCP, and UDP), DoS (types: HTTP, TCP, and UDP), and Reconnaissance (types: OS Fingerprint and ServiceScan), KNN, Random Forest, Naive Bayes, and AdaBoost are used. The entire system is built to withstand a variety of attacks. The BOT- IoT dataset is utilized for the entirety of this study. With the Random Forest approach on this dataset, the results demonstrate effectiveness of the investigation with 99.99% accuracy for the multiclass classification of intrusions. The confusion matrix, classification report are used to analyze the results, in which metrics like precision, recall, and F1-score are considered.