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
46
47
48
49
50
51
52
53
54
55
56
|
Missing data ** New **
XGobi can now handle missing values. In the data file, the missing
values are represented by the character '.' or by the words na or NA.
(Sounds like Sesame Street, no?) To be quite clear, here are three sample
rows:
467 585 na 43 NA
. 580 86 na 12
425 . 77 38 4
______________________ Imputation _________________________
Rescale when imputing
A toggle button that controls whether the view should be rescaled
when a new imputation is performed or read.
Perform random imputation
Use 'group' variable
Instruct XGobi to sample from the present values for each variable
to populate the missing values. If you have done some brushing to
partition the cases, you could turn on the button "Use 'group' variable",
and then the sampling will only be done using cases brushed with the
same color and glyph.
Assign fixed value
10% above max
10% below min
Specify:
An alternative to imputation: assign a fixed value which is either
some fixed percentage above (below) the maximum for each variable, or
specify some fixed value for the missings in the entire data set.
Imputation menu
If you have calculated one or more sets of imputed values that
you would like to view and explore using XGobi, you must construct
two files, datafile.imp (containing the imputed data) and datafile.impnames
(a list of imputation names, one to a line, to be used in the menu).
The structure of the data in datafile.imp is as follows:
Each column is an entire set of imputed data, ordered such that
they will be used to fill in the missing data in column-wise fashion.
Let's use the example above, but start by numbering the missing values:
467 585 na_1 43 na_3
463 580 86 na_2 12
Each column of datafile.imp will contain
imputed_value_1
imputed_value_2
imputed_value_3
|
