Skip to content

Multiple-Response Analysis: Cleaning of Duplicate Codes

5 messages · G.Maubach at weinwolf.de, Bert Gunter, Boris Steipe

Original
G.Maubach at weinwolf.de
# 
Hi All,

in my current project I am working with multiple-response questions 
(MRSets):

-- Coding --
100 Main Code 1
110 Sub Code 1.1
120 Sub Code 1.2
130 Sub Code 1.3

200 Main Code 2
210 Sub Code 2.1
220 Sub Code 2.2
230 Sub Code 2.3

300 Main Code 3
310 Sub Code 3.1
320 Sub Code 3.2

The coding for the variables is to detailed. Therefore I have recoded all 
sub codes to the respective main code, e.g. all 110, 120 and 130 to 100, 
all 210, 220 and 230 to 200 and all 310, 320 and 330 to 300.

Now it happens that some respondents get several times the same main code. 
If the coding was done for respondent 1 with 120 and 130 after recoding 
the values are 100 and 100. If I count this, it would mean that I weight 
the multiple values of this respondent by factor 2. This is not my aim. I 
would like to count the 100 for the respective respondent only once.

Here is my script so far:

# -- cut --

library(expss)

d_sample <-
  structure(
    list(
      c05_01 = c(
        110,
        110,
        130,
        110,
        110,
        110,
        110,
        110,
        110,
        110,
        110,
        999,
        110,
        495,
        160,
        110,
        410
      ),
      c05_02 = c(NA,
                 NA, 120, NA, NA, 150, NA, NA, 170, 160, NA, NA, NA, NA, 
170,
                 NA, 130),
      c05_03 = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, 410,
                 NA, NA, NA, NA, NA, NA, NA),
      c05_04 = c(
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_
      ),
      c05_05 = c(
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_
      )
    ),
    .Names = c("c05_01",
               "c05_02", "c05_03", "c05_04", "c05_05"),
    row.names = c(
      "1",
      "2",
      "3",
      "4",
      "5",
      "10",
      "11",
      "12",
      "13",
      "14",
      "15",
      "20",
      "21",
      "22",
      "23",
      "24",
      "25"
    ),
    class = "data.frame"
  )

c05_xx_r01 <- d_sample %>%
  select(starts_with("c05_")) %>%
  recode(c(
    110 %thru% 195 ~ 100,
    210 %thru% 295 ~ 200,
    310 %thru% 395 ~ 300,
    410 %thru% 495 ~ 400,
    510 %thru% 595 ~ 500,
    810 %thru% 895 ~ 800,
    910 %thru% 999 ~ 900))
names(c05_xx_r01) <- paste0("c05_0", 1:5, "_r01")
d_sample <- cbind(d_sample, c05_xx_r01)

# -- cut --

I would like to eliminate all duplicates codes, e. g. 100 and 100 for 
respondents in row 3, 6, 13, 14 and 15 to 100 only once:

# -- cut --
d_sample_1 <-
  structure(
    list(
      c05_01 = c(
        110,
        110,
        130,
        110,
        110,
        110,
        110,
        110,
        110,
        110,
        110,
        999,
        110,
        495,
        160,
        110,
        410
      ),
      c05_02 = c(NA,
                 NA, 120, NA, NA, 150, NA, NA, 170, 160, NA, NA, NA, NA, 
170,
                 NA, 130),
      c05_03 = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, 410,
                 NA, NA, NA, NA, NA, NA, NA),
      c05_04 = c(
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_
      ),
      c05_05 = c(
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_,
        NA_real_
      ),
      c05_01_r01 = c(
        100,
        100,
        100,
        100,
        100,
        100,
        100,
        100,
        100,
        100,
        100,
        900,
        100,
        400,
        100,
        100,
        400
      ),
      c05_02_r01 = c(NA, NA, NA, NA, NA, NA, NA, NA,
                     NA, NA, NA, NA, NA, NA, NA, NA, 100),
      c05_03_r01 = c(NA, NA,
                     NA, NA, NA, NA, NA, NA, NA, 400, NA, NA, NA, NA, NA, 
NA, NA),
      c05_04_r01 = c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
                     NA, NA, NA, NA, NA, NA),
      c05_05_r01 = c(NA, NA, NA, NA, NA,
                     NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA)
    ),
    .Names = c(
      "c05_01",
      "c05_02",
      "c05_03",
      "c05_04",
      "c05_05",
      "c05_01_r01",
      "c05_02_r01",
      "c05_03_r01",
      "c05_04_r01",
      "c05_05_r01"
    ),
    row.names = c(
      "1",
      "2",
      "3",
      "4",
      "5",
      "10",
      "11",
      "12",
      "13",
      "14",
      "15",
      "20",
      "21",
      "22",
      "23",
      "24",
      "25"
    ),
    class = "data.frame"
  )

# -- cut --

How could I achieve this?

Kind regards

Georg
Bert Gunter
# 
If I understand you correctly, one way is:
[1] "A" "B" "C" "A" "B" "C" "A" "B" "C" "A" "B" "C"
[1] "A" "B" "C"


?duplicated

-- Bert

Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )
On Tue, Apr 25, 2017 at 9:36 AM, <G.Maubach at weinwolf.de> wrote:
Boris Steipe
# 
How about:

d_sample_1 <- floor(d_sample/100) * 100

for (i in 1:nrow(d_sample_1)) {
    d_sample_1[i, duplicated(unlist(d_sample_1[i, ]))] <- NA 
}


B.
G.Maubach at weinwolf.de
# 
Hi Bert,

many thanks for your reply. I appreciate your help a lot.

I would like to do the operation (= finding the duplicates) row-wise.

During this night a solution showed up in my dreams :) Instead of using 
duplicates() to flag and filter the values I could use unique instead with 
the same result. I tested:

# -- cut --

apply(X = c05_xx_r01, MARGIN = 1, unique)

# -- cut --

This finds the unique values for each row. That is nice but lacks the 
requirement that I need a dataframe with a set of variables back that is 
as long as the total amount of unique values for the complete 
data.frame/matrix or the amount of variable of the original data.frame 
respectively.

The result of the above operation gives a list instead of a data.frame due 
to the fact that the amount of resulting values vary from 1 to 7. 
Therefore no data.frame but a list is returned.

I search the web for a solution and found:

http://stackoverflow.com/questions/15753091/convert-mixed-length-named-list-to-data-frame

The complete solution would then look like:

# -- cut --

library(stringi)
library(tidyverse)
my_list <- apply(c05_xx_r01, MARGIN = 1, unique)
my_tibble <- as_tibble(stringi::stri_list2matrix(my_list, byrow = TRUE)
# DONE !

# -- cut --

All-in-all thanks again for your help.

Kind regards

Georg

P.S: I had a look into ?unique. The statement "unique(c05_xx_r01, MARGIN = 
1) does not do the job, cause this looks for unique combinations of values 
on all columns. But that is not the desired outcome.




Von:    Bert Gunter <bgunter.4567 at gmail.com>
An:     G.Maubach at weinwolf.de, 
Kopie:  R-help <r-help at r-project.org>
Datum:  25.04.2017 19:10
Betreff:        Re: [R] Multiple-Response Analysis: Cleaning of Duplicate 
Codes



If I understand you correctly, one way is:
[1] "A" "B" "C" "A" "B" "C" "A" "B" "C" "A" "B" "C"
[1] "A" "B" "C"


?duplicated

-- Bert

Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )
On Tue, Apr 25, 2017 at 9:36 AM, <G.Maubach at weinwolf.de> wrote:
all
code.
I
NA,
http://www.R-project.org/posting-guide.html

  
  


  


      

    

    

      
      

        


  

        

      G.Maubach at weinwolf.de
    

    

      
      #
    

  


  

      
Hi Bert,

many thanks for your reply. I appreciate your help a lot.

I would like to do the operation (= finding the duplicates) row-wise.

During this night a solution showed up in my dreams :) Instead of using 
duplicates() to flag and filter the values I could use unique instead with 
the same result. I tested:

# -- cut --

apply(X = c05_xx_r01, MARGIN = 1, unique)

# -- cut --

This finds the unique values for each row. That is nice but lacks the 
requirement that I need a dataframe with a set of variables back that is 
as long as the total amount of unique values for the complete 
data.frame/matrix or the amount of variable of the original data.frame 
respectively.

The result of the above operation gives a list instead of a data.frame due 
to the fact that the amount of resulting values vary from 1 to 7. 
Therefore no data.frame but a list is returned.

I search the web for a solution and found:

http://stackoverflow.com/questions/15753091/convert-mixed-length-named-list-to-data-frame

The complete solution would then look like:

# -- cut --

library(stringi)
library(tidyverse)
my_list <- apply(c05_xx_r01, MARGIN = 1, unique)
my_tibble <- as_tibble(stringi::stri_list2matrix(my_list, byrow = TRUE)
# DONE !

# -- cut --

All-in-all thanks again for your help.

Kind regards

Georg

P.S: I had a look into ?unique. The statement "unique(c05_xx_r01, MARGIN = 
1) does not do the job, cause this looks for unique combinations of values 
on all columns. But that is not the desired outcome.




Von:    Bert Gunter <bgunter.4567 at gmail.com>
An:     G.Maubach at weinwolf.de, 
Kopie:  R-help <r-help at r-project.org>
Datum:  25.04.2017 19:10
Betreff:        Re: [R] Multiple-Response Analysis: Cleaning of Duplicate 
Codes



If I understand you correctly, one way is:

            

      
      
      
    

        
[1] "A" "B" "C" "A" "B" "C" "A" "B" "C" "A" "B" "C"

            

      
      
      
    

        
[1] "A" "B" "C"


?duplicated

-- Bert

Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )

        
On Tue, Apr 25, 2017 at 9:36 AM,  <G.Maubach at weinwolf.de> wrote:

            

      
      
      
    

        
all

            

      
      
      
    

        
code.

            

      
      
      
    

        
I

            

      
      
      
    

        
NA,

            

      
      
      
    

        

      
      
    

        
http://www.R-project.org/posting-guide.html