Automated model building and evaluation for data mining system

摘要

A method, system, and computer program product that automatically generates data mining models of adequate or even optimum quality in a way that reduces the need for user interaction and reduces the cost and improves the quality of model building. A method for automatically building and evaluating data mining models comprises the steps of receiving information specifying models to be built, generating a model setting array based on the information specifying models to be built, the model setting array including at least one model settings combination, preparing specified source data for training and testing each model to be built, for each model settings combination included in the model settings array, performing the steps of building a model based on the model settings combination, evaluating the built model, and outputting results of the model building and evaluation, selecting a model from among all the built models based on a selection criterion, and outputting the selected model.