Skip to the main content

Original scientific paper

The Algorithm for Discriminative Anlysis in Mahalanobis' Space

Ksenija Bosnar ; Institut za kineziologiju Fakulteta za fizičku kulturu Sveučilišta u Zagrebu
Konstantin Momirović ; Institut za kineziologiju Fakulteta za fizičku kulturu Sveučilišta u Zagrebu
Franjo Prot ; Institut za kineziologiju Fakulteta za fizičku kulturu Sveučilišta u Zagrebu


Full text: croatian pdf 544 Kb

page 31-36

downloads: 147

cite


Abstract

The algorithm was proposed and the program was written for the canonic discriminative analysis of a group of variables with non-singular matrix of co-variance, transformed into the Mahalanobis' form.
The algorithm was proposed and the program for standard algorithms for discriminative analysis in that it determines the discriminative coefficients and discriminative factors in both the metrics of standardized and in the metrics of variables transformed into the Mahalanobis' form: in that it rotates the significant discriminative factors into a position which maximizes the normalized quantimax criterion in which the number of retained factors is determined according to the informational contribution of factors to the explanation of the entire inter-group variance; in that it determines the group centroids for both the unrotated and the rotated discriminative variables in the co-oridnate systems with the starting point in the centroid of the entire sample; in that is statistically cheks the significance of discriminative functions retained by the informational criterion by means of the X2 test (Rao, 1951); in that it cheks the results of the X2 test by means of the single-factor, univariant analysis of variance for each retained rotated and unrotated discriminative variable.

Keywords

discriminative analysis; Mahalanobis, space

Hrčak ID:

250060

URI

https://hrcak.srce.hr/250060

Publication date:

1.10.1984.

Article data in other languages: russian croatian

Visits: 916 *