|
Abstract:
|
Ensemble methods have been demonstrated to be competitive with other machine learning approaches for classification and described for nominal, continuous, and survival responses. In a large number of biomedical applications, the class to be predicted may be ordinal. Examples of ordinal outcomes include TNM stage (I, II, III, IV) and drug toxicity (none, mild, moderate, severe). While nominal methods may be applied to ordinal response data, in so doing some information is lost that may improve the predictive performance of the classifier. As ordinal classification methods have been neglected in the machine learning literature, extensions for predicting an ordinal response using a bootstrap aggregating framework are presented in this talk. These extensions will be evaluated against traditional nominal and continuous ensemble methods using simulated and benchmark datasets.
|
