The selection stage of knowledge discovery in databases (KDD) reduces the project scope to features relevant to the data mining objective while maintaining the final model’s effectiveness. The scope reduction is important because the next stage of KDD, compilation (propositionalization), is a resource-intensive task that converts the data from a relational dataset into a propositional dataset for data mining. However, KDD selection is done manually because conventional feature selection techniques are not compatible with the pre-compiled, relational data. In this chapter, we introduce an automated feature selection before propositionalization (FSbP) and compare its characteristics with traditional feature selection techniques. We provide a use case for FSbP in KDD for oil drilling data, where FSbP is vital for timely project completion. We also briefly discuss the implementation of FSbP algorithms for the above use case, areas for improvement, and limitations of FSbP.