The purpose of this article is to compare and contrast two approaches of assessing the predictive power of an estimated binary dependent-variable classification model, regardless of the modeling method used. One approach is the traditional two-by-two classification table, appropriate for small data settings like clinical experiments. The second approach - the decile table - has become for most modelers a generalized measure of model performance, for a either binary or continuous dependent variable. The decile table is widely used for today's big data. I outline how to construct both tables, and pose questions to raise awareness that each approach has its own weakness
Read more.