Case-Based Reasoning: The Search for Similar Solutions and Identification of Outliers

Abstract

The present paper applies the case-based reasoning (CBR) technique to the problem of outlier detection. Although CBR is a widely investigated method with a variety of successful applications in the academic domain, so far, it has not been explored from an outlier detection perspective. This study seeks to address this research gap by defining the outlier case and the underlining specificity of the outlier detection process within the CBR approach. Moreover, the case-based classification (CBC) method is discussed as a task type of CBR. This is followed by the computational illustration of the approach using selected classification methods, that is, linear regression, distance-based classifier, and the Bayes classifier.