Introducing optimal experimental design in predictive modeling: A motivating example