In this chapter, we bring the context of classification and regression together and look at a common issue: feature ranking and selection. First, we give the motivations, and the usefulness and importance of these feature engineering efforts become obvious. A rich set of various measures for feature ranking are then presented, leading to the next question: how can we search out an optimal subset to achieve the best performance? Some solutions from three approaches are introduced: filter-based, wrapper-based, and embedded methods.