File: VarTransform

package info (click to toggle)
xgobi 19980302-1
  • links: PTS
  • area: non-free
  • in suites: hamm
  • size: 3,344 kB
  • ctags: 4,179
  • sloc: ansic: 48,103; makefile: 843
file content (45 lines) | stat: -rw-r--r-- 1,664 bytes parent folder | download | duplicates (3)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Variable Labels

Clicking left on the variable label within each variable selection
box causes a menu to appear.  This menu lists the available variable
transformations, and the current transformation for the corresponding
variable is highlighted.

If you try to select a transformation that can't be applied to a
particular variable (If, for example, you try to take a square
root of a variable which has some negative values ...), nothing will
be changed and an error message will be printed.

The transformation selected will be applied not only to the selected
variable, but to all members of any group to which that variable
belongs.  Variable grouping occurs upon initialization, and is used to
maintain common scale and transformation among members of the group.
At startup, assuming your data is in a file named 'datafile,' XGobi
looks for a file named datafile.vgroups to gather grouping
information.  To group variables together construct the
datafile.vgroups file to be a single row containing an integer for each
variable in the data matrix.  If variables are to be grouped together
give them the same integer.

It is possible to transform variables during rotation, grand tour, or
in any other mode or action.

Column labels can be supplied in a file named fname.col, where fname
or fname.data is the name of the data file, one label per row.

The current transformations are:

  Restore Variable
  Negative : -1 * x_j
  Inverse : 1 / x_j
  Natural Log
  Fourth Root
  Cube Root
  Square Root
  Square
  Cube
  Fourth Power
  Scale to [0,1]
  Standardize : (x_j - mean(x)) / sigma(x)
  Discretize, 2 levels: x_j > median(x) = 1; x_j <= median(x) = 0