R ggplot2 geom_area() not working

Question

I have a dataset that looks like this:

head():

                          ViolDesc RESULTDTTM count cumsum
1  Adequate Handwashing/Where/When/How 2007-03-05     1      1
2  Adequate Handwashing/Where/When/How 2007-07-09     1      2
3  Adequate Handwashing/Where/When/How 2007-07-16     1      3
4  Adequate Handwashing/Where/When/How 2007-07-17     3      6
5  Adequate Handwashing/Where/When/How 2007-07-23     3      9
6  Adequate Handwashing/Where/When/How 2007-07-27     1     10
7  Adequate Handwashing/Where/When/How 2007-07-30     1     11
8  Adequate Handwashing/Where/When/How 2007-07-31     1     12
9  Adequate Handwashing/Where/When/How 2007-08-07     1     13
10 Adequate Handwashing/Where/When/How 2007-08-13     2     15

dput() head sample of 50 rows:

structure(list(ViolDesc = c("Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How", "Adequate Handwashing/Where/When/How", 
"Adequate Handwashing/Where/When/How"), RESULTDTTM = structure(c(1173052800, 
1183939200, 1184544000, 1184630400, 1185148800, 1185494400, 1185753600, 
1185840000, 1186444800, 1186963200, 1187049600, 1187136000, 1187654400, 
1187827200, 1187913600, 1188172800, 1188432000, 1188864000, 1188950400, 
1189036800, 1189468800, 1189641600, 1189987200, 1190246400, 1190332800, 
1190764800, 1190851200, 1190937600, 1191283200, 1191369600, 1191542400, 
1192406400, 1192492800, 1192665600, 1193011200, 1193270400, 1193702400, 
1193788800, 1193875200, 1193961600, 1194220800, 1194307200, 1194393600, 
1194480000, 1194566400, 1194912000, 1195084800, 1195171200, 1195430400, 
1195516800), class = c("POSIXct", "POSIXt"), tzone = "UTC"), 
    count = c(1L, 1L, 1L, 3L, 3L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 
    1L, 3L, 1L, 1L, 3L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 2L, 1L, 1L, 3L, 1L, 1L, 1L, 2L, 1L, 2L, 3L, 2L, 2L, 
    2L, 1L, 1L, 3L, 3L, 1L, 1L, 2L), cumsum = c(1L, 2L, 3L, 6L, 
    9L, 10L, 11L, 12L, 13L, 15L, 17L, 18L, 19L, 22L, 23L, 24L, 
    27L, 29L, 30L, 31L, 32L, 33L, 34L, 35L, 36L, 37L, 38L, 39L, 
    40L, 42L, 43L, 44L, 47L, 48L, 49L, 50L, 52L, 53L, 55L, 58L, 
    60L, 62L, 64L, 65L, 66L, 69L, 72L, 73L, 74L, 76L)), .Names = c("ViolDesc", 
"RESULTDTTM", "count", "cumsum"), row.names = c(NA, 50L), class = "data.frame")

dput() tail sample of 50 rows:

structure(list(ViolDesc = c("Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's.", 
"Washing fruits and veg's.", "Washing fruits and veg's.", "Washing fruits and veg's."
), RESULTDTTM = structure(c(1342656000, 1344556800, 1345680000, 
1350518400, 1350950400, 1351036800, 1354147200, 1360022400, 1360195200, 
1363305600, 1365638400, 1366070400, 1369872000, 1369958400, 1376265600, 
1376524800, 1383696000, 1385942400, 1388016000, 1389744000, 1396224000, 
1400803200, 1401235200, 1403568000, 1405641600, 1406505600, 1407283200, 
1407801600, 1410739200, 1411603200, 1411689600, 1413849600, 1414368000, 
1415145600, 1416787200, 1417651200, 1418860800, 1420675200, 1424908800, 
1425254400, 1425427200, 1425600000, 1426118400, 1428451200, 1429056000, 
1430438400, 1430870400, 1433376000, 1438214400, 1438819200), class = c("POSIXct", 
"POSIXt"), tzone = "UTC"), count = c(1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L), cumsum = c(92L, 
93L, 94L, 95L, 96L, 97L, 98L, 99L, 100L, 101L, 102L, 103L, 104L, 
105L, 106L, 107L, 108L, 109L, 110L, 111L, 112L, 113L, 114L, 115L, 
116L, 117L, 118L, 119L, 120L, 121L, 122L, 123L, 124L, 125L, 126L, 
127L, 128L, 129L, 130L, 131L, 132L, 134L, 135L, 136L, 137L, 138L, 
139L, 140L, 142L, 143L)), .Names = c("ViolDesc", "RESULTDTTM", 
"count", "cumsum"), row.names = 15178:15227, class = "data.frame")

Column classes:

> sapply(r, class)
$ViolDesc
[1] "character"

$RESULTDTTM
[1] "POSIXct" "POSIXt" 

$count
[1] "integer"

$cumsum

[1] "integer"

There are about 15 different ViolDescs.

If I do ggplot(data=r, aes(x=RESULTDTTM, y=cumsum)) + geom_line(aes(color=ViolDesc)) %>% print

I get:

I simply need this line graph to become a stacked area graph. However, when I do this: ggplot(data=r, aes(x=RESULTDTTM, y=cumsum, fill=ViolDesc)) + geom_area() %>% print, I get:

As you can see, it becomes all spiky and not at all like geom_line(). Does anyone know why this is? I feel like the code is so straightforward that it must be something wrong with the dataframe?

I did some experimenting with random data, and it appears to me that geom_area needs a y-value for each group for each x present in the dataset, because otherwise it doesn't know how to stack. Is that the case in your data? — Heroka, Aug 27 '15 at 16:29
my guess is that your x doesn't match, between your different factors, try switching for nrow(r) — erasmortg, Aug 27 '15 at 16:30
I think I see the problem now: some factors have different dates (x-axis) than other factors, so they can't all be plotted on the same x-axis. I guess when one factor sees a date it doesn't have, it fills it as 0/NA making it drop and causing the spikiness. — Alex Petralia, Aug 27 '15 at 17:15
yeah, exactly. sorry, I just saw this message, you need to ping users in order to notify them(@). I'd subset your rows to only the ones that have the same dates and change to `aes(x=nrow(r),y=cumsum)` or something along those lines — erasmortg, Aug 27 '15 at 17:55

jeremycg · Accepted Answer · 2015-08-27T18:11:01.957

4

Your data has x values which do not always match. ggplot assumes you had a 0 for each variable which has a missing value at one time point but another had a measure.

We can fix this by interpolating the values, using the cumulative maximum for each value at each time point.

Here's how you can do it using dplyr and tidyr.

First load the packages:

library(tidyr)
library(dplyr)
library(ggplot2)

Then read in the data and fix the date format:

z <- read.csv("sample.csv")
z$RESULTDTTM <- as.POSIXct(z$RESULTDTTM, format = "%m/%d/%Y")

Now we want to make the data "wide" - one row for each time point, fill in time points with the cumulative maximum for each column, and then make it "long" again to plot:

z <- z %>% spread(ViolDesc, cumsum) %>% #this makes the data wide
           group_by(RESULTDTTM) %>%     #this makes a group for each time point
           summarise_each(funs(sum(., na.rm = TRUE)), -RESULTDTTM, -count) %>%
           #this line grouped time points together by summing - some time points had two measures
           mutate_each(funs(cummax(.)), -RESULTDTTM) %>% #this takes the cumulative maximum
           gather(var, value, -RESULTDTTM) #this shapes the data back to long

Now we can plot it out:

ggplot(data = z, aes(x = RESULTDTTM, y = value)) +
       geom_area(aes(colour = var, fill = var), position = 'stack')

This gives me this graph:

edited Aug 27 '15 at 18:11

answered Aug 27 '15 at 16:25

jeremycg

24,657
5
63
74

Interesting - I tried this exact code (even before) and it gives me the second plot above (ie. spiky).. Looking into this now. Also what is your ggplot version? – Alex Petralia Aug 27 '15 at 16:35
I've uploaded the data here: http://www.sharecsv.com/s/581ce3a8b4b6eb6e51f6dc2ff999d499/sample.csv. There's no null data in the dataframe so I'm looking at other ways to find out where the issue is. `any(is.na(r))` – Alex Petralia Aug 27 '15 at 16:58
That worked! Your answer was way over my head (I've never used tidyr) but I was able to replicate your picture. Thank you so much, especially for the prompt feedback. – Alex Petralia Aug 27 '15 at 18:21
no problem! all it does it make each variable have a value at each time point. If you do want to follow it through, before each `%>%` you can stop and see the values - ie paste in `z %>% spread(ViolDesc, cumsum)`, `z %>% spread(ViolDesc, cumsum) %>% group_by(RESULTDTTM)` etc. etc. to see what the output is and how each step changes it – jeremycg Aug 27 '15 at 18:24
In fact I am doing that now - this looks useful to learn. Thanks! – Alex Petralia Aug 27 '15 at 18:43

R ggplot2 geom_area() not working

1 Answers1

Linked