User behavior analysis has become an important area of research in modern data analytics and cybersecurity. Organizations collect large volumes of user activity data from digital platforms such as websites, mobile applications, and online services. Analyzing this data helps understand user patterns, detect abnormal behavior, and improve system security. This research presents a machine learning-based approach for analyzing user activity using a synthetic dataset generated with the Faker library. The dataset simulates user actions such as login, purchase, click, and view activities along with attributes including timestamp, duration, location, and device type. Data preprocessing techniques including normalization and categorical encoding are applied before training a Random Forest classifier. The model predicts user behavior categories such as positive, neutral, and negative actions. Experimental results demonstrate that machine learning algorithms can effectively identify behavioral patterns from activity data. Visualization techniques are also used to understand feature importance and behavior patterns. The proposed system can assist organizations in monitoring user activity, improving personalization, and detecting suspicious behavior.
