开发者

summarising multiple non-exclusive dummy variables in R into one variable

开发者 https://www.devze.com 2023-03-09 08:04 出处:网络
I was sent a dataset with multiple dummy variables and other variables as well. Basically what I´d like to do is create summary table with summary.formula from rms. However, I do not know how to crea

I was sent a dataset with multiple dummy variables and other variables as well. Basically what I´d like to do is create summary table with summary.formula from rms. However, I do not know how to create a single variable from the multiple dummy variables and they are not mutually exclusive. Is this at all possible. Of course I could do it creating a table etc, but then I cannot use summary.formula and I´d like the summary.formula output to include just the individual levels of the dummy variables.

edit: to clarify: a & b need to be summarized, but they are not mutually exclusive. Since age is recorded for every row I need to summarize a & b into one variable for it to be used in summary.formula. I´ve edited the code below so that 0 and 1 are changed into NA or a,b respectively.

I´d like the summary.formula output to be something like this:

h<-data.frame(a=sample(c("A",NA),100,replace=T),b=sample(c("B",NA),100,replace=T),age=rnorm(100,50,25),epo=sample(c("Y","N"),100,T))





library(rms)

summary.formula(epo~age####+summary variable of a & b######,method="reverse",data=h)



#-----------------
 Descriptive Statistics by epo

+---------+--------------------------+--------------------------+
|         |N                         |Y                         |
|         |(N=56)                    |(N=44)                    |
+---------+--------------------------+--------------------------+
|age      |31.53434/48.90788/67.69096|28.63689/43.93502/57.81834|
+---------+--------------------------+--------------------------+
|sab : A  |         25% (14)         |         16% ( 7)         |
+---------+--------------------------+------开发者_运维百科--------------------+
|   B     |         27% (15)         |         32% (14)         |
+---------+--------------------------+--------------------------+


Using paste() seems to work acceptably.

h$sab <- paste(h$a, h$b, sep="_")
summary.formula(epo~age+sab,method="reverse",data=h)
#-----------------
 Descriptive Statistics by epo

+---------+--------------------------+--------------------------+
|         |N                         |Y                         |
|         |(N=56)                    |(N=44)                    |
+---------+--------------------------+--------------------------+
|age      |31.53434/48.90788/67.69096|28.63689/43.93502/57.81834|
+---------+--------------------------+--------------------------+
|sab : 0_0|         25% (14)         |         16% ( 7)         |
+---------+--------------------------+--------------------------+
|    0_1  |         27% (15)         |         32% (14)         |
+---------+--------------------------+--------------------------+
|    1_0  |         25% (14)         |         34% (15)         |
+---------+--------------------------+--------------------------+
|    1_1  |         23% (13)         |         18% ( 8)         |
+---------+--------------------------+--------------------------+

Another option might be interaction():

summary.formula(epo~age+interaction(a,b),method="reverse",data=h)

If instead you want a logical 'OR" applied to the combination of variables, then use:

h$a_or_b <- with(h, a|b)
summary.formula(epo ~ age+ h$a_or_b,method="reverse",data=h)
0

精彩评论

暂无评论...
验证码 换一张
取 消