On a Generalization of the Gini Correlation for Statistical Data Mining.

摘要

Cost-sensitive learning has received growing attention in the last twenty years. It is a generalization of the traditional classification problem where costs are taken into consideration instead of simply accuracy. With the rapid development of machine-learning methods in this area, it is of great importance to have proper measures to compare various learners in order to select the optimal one. This thesis develops a new measure by generalizing a commonly used asymmetric measure in social economics, called Gini correlation. The new definition, called the generalized Gini correlation, is found to include special cases that are equivalent to common evaluation measures used in data mining, for example, the LIFT measures for a binary response and the expected profit measure for a monetary response. We consider estimation and inference regarding this generalized Gini correlation. The asymptotic distribution of the estimated correlation is derived with the help of some empirical process theory. We also propose several ways of constructing confidence intervals and demonstrate their performance numerically.