During regular operations, cloud computing environments produce a significant amount of system logs and event recordings. It is challenging and time-consuming to manually monitor these logs in order to identify suspect activity. Conventional security measures frequently rely on rule-based methods that might not be able to identify new or developing cyberthreats. This study suggests a machine learning-based method for identifying questionable system activity from event log data in order to overcome this difficulty. The suggested system looks for unusual behaviour patterns in system activities by analysing attributes including process ID, user ID, event ID, and return values. Before the machine learning model is trained, the dataset is cleaned and prepared using data preparation techniques. System events are divided into two categories by the trained model: suspicious and safe. Python and web technologies are used to provide a web-based interface that enables users to view prediction results and effectively analyse system activities. The suggested method can successfully detect suspicious occurrences and enhance the general security monitoring procedure in cloud settings, according to experimental data. The approach improves the ability to identify possible security risks while reducing the need for manual examination.