With the rapid advancement of digital systems and related information technologies, there is a growing tendency in the global economy to develop digital CRM systems. This research uses a real -world study to predict customer churn and recommends the use of PyCaret to improve a customer churn prediction model. Unlike most studies, this work seeks to employ PyCaret Toolkit, an open source machine learning library designed to make executing typical activities in a machine learning project simple. As a result, a client cluster with a higher risk of fraud has been identified.