oneClass: One-class classification

In benmack/oneClass: One-class classification in the absence of test data

Description

The purpose of the package is to provide i) (an interface to) one-class classifiers, ii) easy access to critical information and diagnostic plots, and iii) an environment for developers to define new methods.

Details

One-class classifiers need to solve a binary classification problem but for the training of the classifier labeled training samples are only available for the class of interest. Model and threshold selection are two critical issues in one-class classification and it is strongly recommended that critical decision of the modelling process are examined carefully by the user.
The package builds upon the powerful caret package (see http://caret.r-forge.r-project.org/). The most important function trainOcc calls train and returns an object of class trainOcc which is a child of the class train. Therefore, many options, e.g. different resampling and pre-processing methods, and tools, e.g. investigation of resampling distributions, from the caret package are also available in the package oneClass.

See Also

The package vignette gives an illustrative introduction to one-class classification and the package oneClass (../doc/oneClassIntro.html)

benmack/oneClass documentation built on Dec. 15, 2020, 7:38 p.m.