What Constitutes a Sufficiently Adequate Binary Classification System?

In this chapter, we analyse the properties of binary classification systems. We use Lagrange’s mean value theorem to establish how its values change as a function of the prevalence threshold and provide both algebraic and geometric definitions of sufficient adequacy. Through the use of the law of cosines we associate different parameters pertaining to a binary classifier such as area under the curve (AUC), central angle, likelihood ratio, and \(\varepsilon \) .