This paper focuses on the identification and elimination of infrequent behaviors in event streams of business processes. By analyzing the event data, we can identify rare behaviors that may indicate errors or inefficiencies in the process. The paper proposes a method for detecting and removing these rare behaviors in order to improve the overall efficiency and accuracy of the business process. The method involves analyzing the frequency and occurrence of behaviors, and using statistical models to predict and identify rare events. This approach has been tested on real-world business processes and has shown promising results in improving process efficiency.