O roli macierzy kowariancji w liniowej funkcji dyskryminacyjnej
THE ROLE OF THE COVARIANCE MATRIX IN A LINEAR DISCRIMINANT FUNCTION

Summary/Abstract: The construction of the well-known and often used in practice linear discrimi-nant function is based on the interclass covariance matrix. In a classic case of discriminant analysis, where there are two populations, linear discriminant function is optimal when both populations have the same covariance matrix.In this article the author describes a problem of using the total covariance matrix calculated for all observations in the discriminant function. As a result of the analysis it is clear that the quality of the classification carried out by using the total covariance matrix and interclass is identical. This interesting phenomenon has be-come an inspiration to find the causes and the geometric interpretation of this rule.

• Details
• Contents