Improved Decision-making in Data Mining

This book explores model selection using decision trees based on discretized data, and the accompanying implications this has for decision-makers. Model selection is a non-trivial exercise with a large impact on decision-making. An information criterion for selecting between competing decision tree models is presented along with a method for quantifying the opportunity costs of choosing a specific decision tree. After the initial information criterion development a real world example from an American insurance company call center is presented. The example includes the development of a program to automate data discretization, decision tree creation, and decision tree selection. A decision tree is chosen and critically analyzed from a managerial decision-making point of view. Actual call center performance data is used as input data and results are identified and presented in a way that is advantageous to managers.