The variance of the error rates of classification rules

✍ Scribed by Terence J. O'Neill

• Book ID: 103930183
• Publisher: Elsevier Science
• Year: 1986
• Tongue: English
• Weight: 519 KB
• Volume: 12
• Category: Article
• ISSN: 0898-1221
• DOI: 10.1016/0898-1221(86)90080-5

✦ Synopsis

Abstractwln this paper an expansion for the variance of the error rate of a classification rule is derived. It is explicitly evaluated for normal distributions and linear rules and also for the quadratic discrimination rule. The results are used to compare the estimates of Fisher's linear (F). the quadratic (Q) and the logistic (£) discriminant rules. It is found that although P is often preferable in terms of expected error rule. it is sometimes marginally more variable.