Re-order y-axis by Number of Observations

Question

I'm trying to re-order a categorical/non-numeric y-axis by the number of observations associated with each value.

I import a table that looks like this:

View(ranges)

ID  | Time | Value | obs
B1  | 1:11 | 1     | 91
B1  | 1:12 | 2     | 91
...
U30 | 2:22 | 99    | 101

Then reorder:

ranges <- ranges[with(ranges, order(obs, ID)), ]

Then plot:

p <- ggplot(ranges, aes(x=Time, y=ID, fill=Value))
p <- p + geom_tile(colour = "white")
p

But the order on the y-axis does not reflect the number observations per ID.

How can I sort the y-axis to reflect the frequency of the ID in the dataframe? Thanks kindly.

Sample data, comma delimited:

ID,DATE,VALUE,OBS
B9148,41275,0133,7
B9148,41276,3344,7
B9148,41277,7907,7
B9148,41278,2052,7
B9148,41279,6617,7
B9148,41280,9411,7
B9148,41281,6302,7
U1821,41275,7585,11
U1821,41276,5086,11
U1821,41277,5731,11
U1821,41278,5358,11
U1821,41279,9977,11
U1821,41280,1446,11
U1821,41281,2728,11
U1821,41282,5369,11
U1821,41283,8171,11
U1821,41284,5817,11
U1821,41285,0166,11
F7484,41275,8971,10
F7484,41276,5635,10
F7484,41277,8629,10
F7484,41278,5832,10
F7484,41279,4775,10
F7484,41280,4519,10
F7484,41281,1527,10
F7484,41282,7861,10
F7484,41283,5713,10
F7484,41284,0824,10
S3804,41275,0573,4
S3804,41276,0123,4
S3804,41277,8468,4
S3804,41278,6258,4
B8088,41275,9117,7
B8088,41276,8489,7
B8088,41277,0902,7
B8088,41278,8684,7
B8088,41279,7229,7
B8088,41280,3587,7
B8088,41281,0907,7

Perhaps use `reorder()`, as demonstrated e.g. [here](http://stackoverflow.com/questions/9741600/sort-boxplot-by-mean-and-not-median-in-r/9741678#9741678). — Josh O'Brien, Nov 21 '13 at 17:25
Sort them by the new column you created and then plot then as normal. — CCurtis, Nov 21 '13 at 19:54
That doesn't seem to work. I'll edit with more specific code. — user1893148, Nov 21 '13 at 20:03
Your data is quite specific. I could generate it artificially, but that would take longer than answering the question itself. In other words, provide some data. — tonytonov, Nov 22 '13 at 08:01

score 1 · Accepted Answer · edited May 23 '17 at 12:05

1

You can do this either manually:

ranges$ID <- factor(ranges$ID, levels=c("S3804", "B8088", "B9148", "F7484", "U1821"))

or more or less automatically:

ranges$ID <- factor(ranges$ID, levels=as.character(unique(ranges$ID)))

The basic idea of why this happens is because factor levels are stored separately and are not affected by ordering, whereas ggplot plots level by level.

You can also refer to this question for a demo on this concept.

Is this what you wanted? Note that if you want reverse order, just add rev() before as.character().

edited May 23 '17 at 12:05

Community

1
1

answered Nov 27 '13 at 13:44

tonytonov

25,060
16
82
98

Exactly what I wanted. Thanks kindly. – user1893148 Nov 28 '13 at 00:42
And for posterity, the key is that ggplot requires you re-order both values and levels. – user1893148 Nov 28 '13 at 01:18

Re-order y-axis by Number of Observations

1 Answers1