frequency table-visualization for complex categorical variables

in the future, please provide R code to re-create some example data :)
read
http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-examplefor
more detail..



# create a data table with three unique columns' values..
# treat these values just like letters
x <-
     cbind(
         sample( LETTERS[1:6] , 100 , replace = TRUE ) ,
         sample( LETTERS[1:6] , 100 , replace = TRUE ) ,
         sample( LETTERS[1:6] , 100 , replace = TRUE )
     )

# look at x.. this is good data i hope?
x

# convert this to a matrix
y <- as.matrix( x )

# i don't think you care about ordering, so sort left-to-rightwards
z <- apply( y , 2 , sort )

# look at your results
z

# paste these results together across the matrix
w <- apply( z , 1 , paste0 , collapse = "" )

# count the final distinct results
table( w )




On Mon, Feb 25, 2013 at 1:04 PM, Niklas Fischer
<niklasfischer980 at gmail.com>wrote:

Dear R users,

I have three questions measuring close relationships.
The questions are same and the respondents put the answer in order.

I'd like to examine the pattern of answers and visualize it.

For example q1 (A,B,C,D,E) and q2 and q3 are the same. If the respondents
selects A B C (so BCA or BAC or CBA or CAB), I'd like to construct
frequency table for ABC and other combinations for example DEF.


Unfortunately, there are many answers, and three-way contingency table
includes lots of cells which make it diffucult to interpret and requires
lots of extra work to organize data.

What is the best way to construct fruequency table of these kind of
variables and to visulize the results with the most simple form


All the bests,
Niklas

         [[alternative HTML version deleted]]

	[[alternative HTML version deleted]]

Hi again,



Thanks for Anthony about the links on reproducible codes.



Thanks for Rui about ordering when rows are intact.



One more question





Here is your code.



x <-
     cbind(
         sample( LETTERS[1:6] , 100 , replace = TRUE ) ,
         sample( LETTERS[1:6] , 100 , replace = TRUE ) ,
         sample( LETTERS[1:6] , 100 , replace = TRUE )
     )

y <- as.matrix( x )

w2 <- apply( y , 1 , paste0 , collapse = "" )
table(w2)







Do you know any trick to organize merge certain elements together?

For example, if the final table includes

BCC, CCB, CBC how should I sum frequency of one element like BCC? I have a
very long table it would be indeed very useful!





Niklas.

2013/2/25 Rui Barradas <ruipbarradas at sapo.pt>

Hello,

I disagree with the way you've sorted the matrix, like this all A's become
first, then B's, etc, irrespective of the respondents. Each row is a
respondent, and the rows should be kept intact, but with a different
ordering. To this effect, use order():

z <- y[order(y[,1], y[,2], y[,3]), ]


Then use the rest of your code.

Or, which would save us the sorting, paste the rows elements together
directly from matrix 'y' and use the fact that table() sorts its output.

w2 <- apply( y , 1 , paste0 , collapse = "" )
table(w2)


Hope this helps,

Rui Barradas

Em 25-02-2013 18:32, Anthony Damico escreveu:

  in the future, please provide R code to re-create some example data :)

read
http://stackoverflow.com/**questions/5963269/how-to-make-**
a-great-r-reproducible-**examplefor<http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-examplefor>
more detail..



# create a data table with three unique columns' values..
# treat these values just like letters
x <-
      cbind(
          sample( LETTERS[1:6] , 100 , replace = TRUE ) ,
          sample( LETTERS[1:6] , 100 , replace = TRUE ) ,
          sample( LETTERS[1:6] , 100 , replace = TRUE )
      )

# look at x.. this is good data i hope?
x

# convert this to a matrix
y <- as.matrix( x )

# i don't think you care about ordering, so sort left-to-rightwards
z <- apply( y , 2 , sort )

# look at your results
z

# paste these results together across the matrix
w <- apply( z , 1 , paste0 , collapse = "" )

# count the final distinct results
table( w )




On Mon, Feb 25, 2013 at 1:04 PM, Niklas Fischer
<niklasfischer980 at gmail.com>**wrote:

  Dear R users,

I have three questions measuring close relationships.
The questions are same and the respondents put the answer in order.

I'd like to examine the pattern of answers and visualize it.

For example q1 (A,B,C,D,E) and q2 and q3 are the same. If the respondents
selects A B C (so BCA or BAC or CBA or CAB), I'd like to construct
frequency table for ABC and other combinations for example DEF.


Unfortunately, there are many answers, and three-way contingency table
includes lots of cells which make it diffucult to interpret and requires
lots of extra work to organize data.

What is the best way to construct fruequency table of these kind of
variables and to visulize the results with the most simple form


All the bests,
Niklas

          [[alternative HTML version deleted]]

         [[alternative HTML version deleted]]