visualize associations between two groups of data

Question

Where each datapoint has a pairing of A and B and there multiple entries in A and multiple entires in B. IE multiple syndromes and multiple diagnoses, although for each datapoint there is one single syndrome-diagnoses pair.

Examples, suggestions, or ideas much appreciated

here's what the data is like. And I want to see connections between values of A and B (how many GG's are linked to TTs etc). Both are nominal datatypes.

ID,A ,B 
1,GG,TT
2,AA,SS
3,BB,XX
4,DD,SS
5,DD,TT
6,CC,XX
7,HH,ZZ
8,AA,TT
9,CC,RR
10,DD,ZZ
11,AA,XX
12,AA,TT
13,DD,SS
14,DD,XX
15,AA,YY
16,CC,ZZ
17,FF,SS
18,FF,XX
19,BB,VV
20,GG,VV
21,GG,SS
22,AA,RR
23,AA,TT
24,AA,SS
25,CC,VV
26,CC,TT
27,FF,RR
28,GG,UU
29,CC,TT
30,BB,ZZ
31,II,TT
32,FF,RR
33,BB,SS
34,GG,YY
35,FF,RR
36,BB,VV
37,II,RR
38,CC,YY
39,FF,VV
40,AA,XX
41,AA,ZZ
42,GG,VV
43,BB,UU
44,II,UU
45,II,SS
46,DD,SS
47,AA,UU
48,BB,VV
49,GG,TT
50,BB,TT

Answer 1

This is what I do. A darker colour indicates a more important combination of A and B.

dataset <- data.frame(A = sample(LETTERS[1:5], 200, prob = runif(5), replace = TRUE), B = sample(LETTERS[1:5], 200, prob = runif(5), replace = TRUE))
Counts <- as.data.frame(with(dataset, table(A, B)))
library(ggplot2)
ggplot(Counts, aes(x = A, y = B, fill = Freq)) + geom_tile() + scale_fill_gradient(low = "white", high = "black")

Or if you prefer lines

library(ggplot2)
dataset <- data.frame(A = sample(letters[1:5], 200, prob = runif(5), replace = TRUE), B = sample(letters[1:5], 200, prob = runif(5), replace = TRUE))
Counts <- as.data.frame(with(dataset, table(A, B)))
Counts$X <- 0
Counts$Xend <- 1
Counts$Y <- as.numeric(Counts$A)
Counts$Yend <- as.numeric(Counts$B)
ggplot(Counts, aes(x = X, xend = Xend, y = Y, yend = Yend, size = Freq)) +
geom_segment() + scale_x_continuous(breaks = 0:1, labels = c("A", "B")) + 
scale_y_continuous(breaks = 1:5, labels = letters[1:5])

This third options add labels to the data points using geom_text().

library(ggplot2)
dataset <- data.frame(
    A = sample(letters[1:5], 200, prob = runif(5), replace = TRUE), 
    B = sample(LETTERS[20:26], 200, prob = runif(7), replace = TRUE)
)
Counts <- as.data.frame(with(dataset, table(A, B)))
Counts$X <- 0
Counts$Xend <- 1
Counts$Y <- as.numeric(Counts$A)
Counts$Yend <- as.numeric(Counts$B)
ggplot(Counts, aes(x = X, xend = Xend, y = Y, yend = Yend)) + 
geom_segment(aes(size = Freq)) + 
scale_x_continuous(breaks = 0:1, labels = c("A", "B")) + 
scale_y_continuous(breaks = -1) + 
geom_text(aes(x = X, y = Y, label = A), colour = "red", size = 7, hjust = 1, vjust = 1) + 
geom_text(aes(x = Xend, y = Yend, label = B), colour = "red", size = 7, hjust = 0, vjust = 0)

Answer 2

Thanks! I think that the connectivity between elements in each class is best visualized by the link graph examples given by both Jonathon and Thierry. Thierry's 2nd which shows the magnitude is definitely where i will start.

update thanks everyone for you ideas and tips!

I came acrossthe bipartite package that has functions to visualize this kind of data. I think its a clean visualization of the relationships I am trying to show.

did:

    library(bipartite)
    dataset <- data.frame(
         A = sample(letters[1:5], 200, prob = runif(5), replace = TRUE), 
         B = sample(LETTERS[20:26], 200, prob = runif(7), replace = TRUE)
     )
    datamat <- as.matrix(table(dataset$A, dataset$B))
    visweb(datamat, text = "interaction", textsize = .8)

giving: visweb result

couldnt put image in as a new user :(