Parameter Tuning using the Out-of-Bootstrap Generalisation Error Estimate for Stochastic Discrimination and Random Forests

Stochastic discrimination is a machine learning algorithm with strong theoretical underpinnings and good published results on UCI datasets. However, it has not been popular amongst practitioners. We look at some of the issues involved in its use, propose the out-of-bootstrap error estimator as a means of tuning stochastic discrimination´s and other classifiers´ performance and contrast stochastic discrimination´s utility with that of a related classification technique of random forests