Real-world data are often prepared for purposes other than data mining and machine learning and, therefore, are represented by primitive attributes. When data representation is primitive, preprocessing data before looking for patterns becomes necessary. If lack of domain experts prevents the use of highly informative attributes, patterns are hard to uncover due to complex attribute interactions. This article suggests a new use of MFE3/GA to restructure the primitive data representation by means of capturing and compacting hidden information into new features in order to highlight them to the learner. Empirical results on Poker Hand data set show that the new use successfully improves learning this concept by means of data reduction, generation of a smaller decision tree classifier, and accuracy improvement.