对角线以下的绘图值重新排序？

Question

我有一个按行组织的数据文件，带有 x 标签、y 标签，然后是一个值。基本结构如下所示:

Factor_1,Factor_2,Number<br/>
apple,apple,1<br/>
banana,apple,1<br/>
apple,kiwi,6<br/>
apple,pear,1<br/>
watermelon,apple,8<br/>
banana,banana,3<br/>
banana,kiwi,2<br/>
banana,pear,1<br/>
banana,watermelon,9<br/>
kiwi,kiwi,9<br/>
pear,kiwi,4<br/>
kiwi,watermelon,4<br/>
pear,pear,3<br/>
pear,watermelon,9<br/>
watermelon,watermelon,1<br/>
...<br/>
...

使用此数据，我正在使用以下代码构建类似于相关矩阵的东西:

library(ggplot2)
library(reshape2)

d <- read.csv("my_file", head=TRUE, sep="\t")
x <- dcast(d, Factor_1~Factor_2)
x.m <- melt(x)
x.m <- ddply(x.m, .(variable))
(p <- ggplot(x.m, aes(variable, Factor_1)) + geom_tile(aes(fill = value), colour = "black") + scale_fill_gradient(low = "black", high = "green")

上面的代码给了我一个看起来像这样的情节:

如何重新排序数据以构建具有以下格式的图表，其中所有数据都分组在对角线下方？

   A  B  K  P  W
W              1
P           3  9
K        9  4  4
B     3  2  1  9
A  1  1  6  1  8

Answer 1

我称它为“水果”:

fruit.tbl <- xtabs(V3 ~ V1+V2, data=fruits)
> melt(fruit.tbl)
           V1         V2 value
1       apple      apple     1
2      banana      apple     0
3        kiwi      apple     0
4        pear      apple     0
5  watermelon      apple     0
6       apple     banana     1
7      banana     banana     3
8        kiwi     banana     0
9        pear     banana     0
10 watermelon     banana     0
11      apple       kiwi     6
12     banana       kiwi     2
13       kiwi       kiwi     9
14       pear       kiwi     0
15 watermelon       kiwi     0
16      apple       pear     1
17     banana       pear     1
18       kiwi       pear     4
19       pear       pear     3
20 watermelon       pear     0
21      apple watermelon     8
22     banana watermelon     9
23       kiwi watermelon     4
24       pear watermelon     9
25 watermelon watermelon     1

mfruit <- melt(fruit.tbl)
is.na(mfruit$value) <- mfruit$value==0

# Needed to swap x and y to get it the way you wanted
 (p <- ggplot(melt(mfruit), aes(V2, V1,fill = value)) + 
                geom_tile( colour = "black") + 
                scale_fill_gradient(low = "black", high = "green")
   )