% @(#)mda_tabl.hlq 17.1.1.1 (ESO-IPG) 01/25/02 17:16:32 %++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ %.COPYRIGHT (c) 1990 European Southern Observatory %.IDENT mda_tab.hlq %.AUTHOR FM, IPG/ESO %.KEYWORDS MIDAS, help files, MDA/TAB %.PURPOSE On-line help file for the command: MDA/TAB %.VERSION 1.0 28-AUG-1986 : Creation, FM %---------------------------------------------------------------- \se SECTION./TABL \es\co MDA/TAB 28-AUG-1986 FM \oc\su MDA/TAB input_table output_table eigenvectors discriminant analysis \us\pu Purpose: Multiple Discriminant Analysis, Discriminant Factor Analysis, Canonical Discriminant Analysis. \up\sy Syntax: MDA/TAB input_table output_table eigenvectors \ys\pa input_table = input table name. \ap\pa output_table = output table name. \ap\pa eigenvectors = output table name (defaults to none). \ap\no Note: Singularities, caused by linear dependance, can be circumvented by first carrying out a PCA, and using a small number of principal components as input to LDA. Singularity problems may arise in matrix inversion, due to very small values: if you get an IERR=2 message, multiplying the input data by an arbitrary constant may painlessly get around the difficulty. \on\exs Examples: \ex MDA/TAB INTAB OUTAB \xe \sxe