Effect of model specification in pattern classification

Date

1979

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

In practical application of statistical pattern recognition techniques, statistical models are often postulated without performing any exploratory data analysis or statistical testing for model validation. As a consequence, the anticipated classification probabilities may appreciably differ from the true classification probabilities. Estimates of proportion of component statistical populations in a mixture sample are also affected. The effect of model misspecification in pattern recognition and pattern recognition related techniques have been examined in this paper.

Description

Keywords

Citation