On the 31st of this month, GPMC will host Professor Rodrigo Heldt (PUCRS) at its weekly research seminar, where he will discuss the topic “Improving churn prediction performance in retail banking.” The conversation will begin at 9 AM and is expected to last one hour. Interested participants can attend in person in the CEPA room at EA or via Zoom. The entire community is welcome!
Now, a description of the project: Predicting customer churn becomes even more challenging when churn events are extremely rare, making it difficult for machine learning algorithms to learn effectively. To overcome this challenge, the research to be presented proposes a meticulous data preprocessing process that combines feature engineering with the use of different machine learning techniques aimed at enhancing the predictive performance of churn prediction models. The proposed framework, empirically applied in a retail bank involving over 3 million customers and approximately 170 million transactions, demonstrated high predictive performance in identifying which customers are likely to churn in the coming months, surpassing alternative methods traditionally used in the literature.
Rodrigo Heldt is an assistant professor in the Graduate Program in Administration at the Pontifical Catholic University of Rio Grande do Sul. He holds a postdoctoral degree in marketing and a PhD in Marketing from the Federal University of Rio Grande do Sul, including a period as a visiting scholar at Texas Christian University (2020). He earned a Master’s in Marketing from the Federal University of Rio Grande do Sul (2014) and a degree in Administration (2011). He is a member of the research project on Customer Management with Banrisul Bank. His research interests include marketing analytics, customer analytics, customer lifetime value, and brand equity.
Meeting link: https://mit.zoom.us/j/92825888029
