Classification Algorithm Validation
Asoke Kumar Nandi, Hosameldin Ahmed · 2019
Classification is an essential task in the machine fault diagnosis framework for assigning class labels, i.e. machine health conditions, to new vibration signal examples. Model selection and performance estimation are the two main tasks considered when dealing with classification problems. These tasks can be achieved by using model-validation techniques. This chapter describes different validation techniques that can be used to evaluate and verify the classification model's performance before proceeding to its application and implementation in machine fault diagnosis problems. The most common approach to cross-validation is data-splitting, which can be done using one of the following techniques: the hold-out technique, random subsampling, k-fold cross-validation, leave-one-out cross-validation, and bootstrapping. Furthermore, various measures can be used to evaluate the performance of a classification model, including overall classification accuracy, a confusion matrix, the receiver operating characteristic, precision, and recall.