This book is the first of its kind to discuss error estimation with a modelbased approach. From the basics of classifiers and error estimators to distributional and Bayesian theory it covers important topics and essential issues pertaining to the scientific validity of pattern classification.