Classification is a term used for the process of assigning an observed unit to one of a small number of labelled groups (classes), such as ‘ordinary’ and ‘unusual or ‘positive’, ‘neutral’ and ‘negative’. The groups are exclusive and exhaustive—a single group is appropriate for every unit. Common applications of classification arise in medical screening, educational tests and licencing examinations, fraud detection and when searching for units with exceptional attributes. The groups may be well defined a priori, or their definition is based on the analysis of a collection of observed units. The term ‘misclassification’ is used for assigning a unit to an inappropriate group, a group to which the unit does not belong. We deal with the setting of two groups of units, called positives and negatives, in which there are two kinds of inappropriate assignments; the corresponding (misclassified) units are called false positives and false negatives. Our task is to minimise the expected loss associated with such misclassification.
Keywords
- False Negative
- Loss Function
- Beta Distribution
- Gray Zone
- Expected Loss
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.