Skip to content

Plotting graph for Missing values

12 messages · Shreyasee, jim holtman, Bart Joosen +1 more

#
What does you data look like?  You could use 'split' and then examine
the data in each range to count the number missing.  Would have to
have some actual data to suggest a solution.
On Sun, Jan 25, 2009 at 8:30 PM, Shreyasee <shreyasee.pradhan at gmail.com> wrote:

  
    
#
Here is an example of how you might approach it:
200605 200606 200607 200608 200609 200610 200611 200612 200701 200702 200703
    21     22     16     18     16     15     16     17     14     16     13

        
On Sun, Jan 25, 2009 at 8:51 PM, Shreyasee <shreyasee.pradhan at gmail.com> wrote:

  
    
#
YOu can save the output of the tapply and then replicate it for each
of the variables.  The data can be used to plot the graphs.
On Sun, Jan 25, 2009 at 9:38 PM, Shreyasee <shreyasee.pradhan at gmail.com> wrote:

  
    
#
do:

str(dos)
str(patientinformation1)

They must be the same length for the command to work: must be a one to
one match of the data.
On Sun, Jan 25, 2009 at 10:23 PM, Shreyasee <shreyasee.pradhan at gmail.com> wrote:

  
    
#
I would say that "pat1" and "dos" were not of the same length.  Check
your code and objects to verify this; that is what the error message
is saying.  You said you added the "patientinformation1" variable, but
it does not seem to appear in the error message.
On Sun, Jan 25, 2009 at 11:48 PM, Shreyasee <shreyasee.pradhan at gmail.com> wrote:

  
    
#
seems like you added patientinformation1, but still use pat1 in the tapply
call.

Bart
#
Hi Jim

r-help-bounces at r-project.org napsal dne 26.01.2009 15:44:32:
generate
You are really patient. I presume Shreyasee does not know much about data 
structures and function use in R. It probably could help a lot if s/he 
looked into same basic documents like R intro.

If I understand correctly what was done is

pat1 <- rbinom(length(patientinformation1), 1, .5)

what does not make much sense as it code an artificial data as well and 
most probably there is "dos" version in memory which was constructed 
during testing your code and which has length 335. This could result in 
mentioned error
Then note
the ASC
if str(ds) is issued, it could reveal what kind of data s/he has. 
Also format(dos, ...) would not work as dos is factor not Date
If it was
Group.1 pat1 pat2
1   200605   12   16
2   200606   20   18
3   200607   12   13
4   200608   18   15
5   200609   18   11
6   200610   17   15
7   200611   19   17
8   200612   14   15
9   200701   14   18
10  200702   13   13
11  200703   16   19

could do the trick if patientinformation variables had the same structure 
as you anticipate which is not true
Well, if Shreyasee manage to redefine dos to Date mode (which will not be 
straightforward if "dos" has awkward structure), then something like

aggregate(ds[,-1], list(format(ds$dos, "%Y%m")), function(x) sum(x==""))

could do the trick.

Regards
Petr
<shreyasee.pradhan at gmail.com> wrote:
the ASC
wrote:
to
<shreyasee.pradhan at gmail.com>
for
:
each
the
for
<jholtman at gmail.com>
by='1
values
16
the
patientinformation3)
<jholtman at gmail.com>
then
have
each
missing
reproducible
http://www.R-project.org/posting-guide.html