Integrating RapidMiner in Business Analytics Education: An Instructional Approach for Skill Development

Abstract

The continued growth of business analytics discipline raises the need for students as future professionals to be trained in business analytics concepts and applications to enable data-driven decision-making within organizations. As business analytics evolves to incorporate data mining and machine learning applications, students need to develop an overall understanding of the process of acquiring, preparing, and analyzing data. This paper describes a framework involving five stages—preparation, exploration, modeling, optimization, and validation—that can be used to instruct students on the business analytics process in the context of the RapidMiner software package. Further, this paper illustrates the application of the framework using a specific example that uses decision trees along with a discussion of the descriptive statistics, visual analysis using charts, identifying the variables for analysis, ways to optimize and validate models, and assess model performance. This includes training and testing (holdout) samples, unbalanced data, confusion matrix, precision and recall metrics, and AUC and F1 metrics. Student performance on individual assignments following in-class instruction and demonstration based on the framework shows that it was helpful in student learning. Although the framework was tested in the context of RapidMiner, it can be extended for business analytics instruction using other software tools.