next up previous contents
Next: References Up: Multivariate Analysis Methods Previous: Correspondence Analysis

Related Table Commands

A tutorial command (TUTORIAL/MVA) may be used to see the very straightforward way in which the methods described here are used. Help commands may also be used for the required syntax.

A few MIDAS Table commands, of particular interest for the use of multivariate routines, are mentioned here for convenience. Note that the multivariate routines use all columns and rows in the input tables, and hence tables must first be ``rearranged'', as desired, before analysis. MIDAS subsequently offers a powerful environment for plotting results.

To show the characteristics of table mytab, to read its values to the screen, to print it out on a good quality device, to delete column 4 of mytab, and to construct table mytab2 from mytab, with three columns (PC1, PC2 and PC3) in the former:

SHOW/TAB mytab
READ/TAB mytab
PRINT/TAB mytab
DELE/TAB mytab
$\char93 4$
PROJ/TAB mytab mytab2 :PC1 :PC2 :PC3

To select group 2 members, defined in column 4, from table MYTAB; then plot coordinates defined in columns 1 and 2; then select group 3 members; overlay the coordinates of these points on the initial plot with a different symbol; and finally to get a good quality hard-copy representation:

ASSIGN/PLOT LASER NOSPOOL
SELECT/TAB mytab
$\char93 4$.EQ.2
SET/PLOT STYPE=2
PLOT/TAB mytab
$\char93 1$$\char93 2$
SELE/TAB mytab ALL
SELE/TAB mytab
$\char93 4$.EQ.3
SET/PLOT STYPE=6
OVER/TAB mytab
$\char93 1$$\char93 2$
SEND/PLOT LASER

Since the scales of group 3 members, may differ from the scale of group 2 members in the above, it may be advisable to issue a SET/PLOT XAXIS= YAXIS= command, with attendant minima and maxima for the axis, before plotting.

For further details of parameters to these commands, and for other relevant commands, the Tables chapter in this manual should be consulted.

Also in this Tables chapter are some routines for regression.


next up previous contents
Next: References Up: Multivariate Analysis Methods Previous: Correspondence Analysis
Petra Nass
1999-06-15