Plotting rectangles in ggplot2 - Invalid input: time_trans works with objects of class POSIXct only

Question

I have this data

df <- structure(list(IndID = structure(c(16L, 15L, 14L, 13L, 12L, 11L, 
10L, 9L, 8L, 7L, 6L, 5L, 4L, 3L, 2L, 1L), .Label = c("16", "15", 
"14", "13", "12", "11", "10", "9", "8", "7", "6", "5", "4", "3", 
"2", "1"), class = "factor"), StartDate = structure(c(1313042400, 
1312956000, 1313560800, 1363672800, 1374040800, 1374040800, 1374040800, 
1374040800, 1374040800, 1374040800, 1374040800, 1365832800, 1365919200, 
1366178400, 1395727200, 1395813600), class = c("POSIXct", "POSIXt"
)), EndDate = structure(c(1377928800, 1378015200, 1378015200, 
1386572400, 1410760800, 1410760800, 1410760800, 1410674400, 1410760800, 
1406959200, 1399356000, 1427868000, 1394517600, 1428213600, 1428040800, 
1420959600), class = c("POSIXct", "POSIXt"))), .Names = c("IndID", 
"StartDate", "EndDate"), row.names = c(NA, -16L), class = "data.frame")

  IndID  StartDate    EndDate
1     1 2011-08-11 2013-08-31
2     2 2011-08-10 2013-09-01
3     3 2011-08-17 2013-09-01
4     4 2013-03-19 2013-12-09
5     5 2013-07-17 2014-09-15
6     6 2013-07-17 2014-09-15

and can make this plot

    library(lubridate)
    require(gglopt2)
    df$IndID <- factor(df$IndID, levels = rev(df$IndID))



 p1 <- ggplot(df, aes(x=IndID, fill = IndID))+
      geom_rect(aes(x = IndID, xmin = as.numeric(IndID) - 0.45, xmax = as.numeric(IndID) + 0.45, ymin = StartDate, ymax = EndDate))+
      coord_flip()+
      xlab("IndID")+
      ylab("Time")+
      ggtitle("Individual Monitoring Periods")+
      geom_text(aes(y = StartDate + as.difftime(8, unit = "weeks"), label = paste(month(StartDate, label = T), year(StartDate), sep = "-"))) +
      geom_text(aes(y = EndDate - as.difftime(9, unit = "weeks"), label = paste(month(EndDate, label = T), year(EndDate), sep = "-")))

    p1

I addition, I want to shade the region between June 19 and October 19 for every year. To do so, I made a data.frame of dates which were then converted to POSIXct format in a new data.frame. (yes, this is clunky..)

temp <- data.frame(
    start = as.Date(c('2011-06-19', '2012-06-19', '2013-06-19', '2014-06-19', '2015-06-19')), 
    end   = as.Date(c('2011-10-19', '2012-10-19', '2013-10-19', '2014-10-19', '2015-10-19')))
str(temp)
dateRanges <- data.frame(
    start = as.POSIXct(temp [,1], "%Y-%m-%d") + hours(6),
    end   = as.POSIXct(temp [,2], "%Y-%m-%d") + hours(6))
str(dateRanges)

When I try to add the new rectangles to the plot with the following code, i get the error indicated in the post title.

p1 + geom_rect(data = dateRanges, aes(xmin = start , xmax = end, ymin = -Inf, ymax = Inf), inherit.aes= F, alpha = 0.4, fill = c("lightblue"))

As far as I can tell by looking at the str() of the dateRanges, they are correctly formatted as POSIXct class.

I did see the similar post here but am still struggling to connect the dots between the to related issues and fix my problem. Thanks in advance for any suggestions.

Maybe you should add the tag 'gantt-chart' or just mention the term somewhere, that might help for future searches. — maj, Sep 15 '15 at 20:32

jeremycg · Accepted Answer · 2015-09-15T20:18:08.287

Your problem is you ran coord_flip() in your initial call. The plot is flipped, but ggplot still thinks that the x and y are the original x and y.

So, to fix, just switch your x and y aes in the final geom_rect:

p1 + geom_rect(data = dateRanges, aes(ymin = start , ymax = end, xmin = -Inf, xmax = Inf), inherit.aes= F, alpha = 0.4, fill = c("lightblue"))
p1

EDIT: To make the bars behind requires a little bit of fiddling. first we have to call the geom_rect which makes the bars so it is behind, and we move the xmin and xmax out of the aes call to avoid messing with the axis and levels of factors:

p1 <- ggplot(df, aes(x=IndID, fill = IndID))+
  geom_rect(data = dateRanges, aes(ymin = start, ymax = end), xmin = -Inf, xmax = Inf, alpha = 0.4, inherit.aes=FALSE, fill = "lightblue")+
  geom_rect(data = df, aes(x = IndID, xmin = as.numeric(IndID) - 0.45, xmax = as.numeric(IndID) + 0.45, ymin = StartDate, ymax = EndDate))+
  coord_flip()+
  xlab("IndID")+
  ylab("Time")+
  ggtitle("Individual Monitoring Periods")+
  geom_text(aes(y = StartDate + as.difftime(8, unit = "weeks"), label = paste(month(StartDate, label = T), year(StartDate), sep = "-"))) +
  geom_text(aes(y = EndDate - as.difftime(9, unit = "weeks"), label = paste(month(EndDate, label = T), year(EndDate), sep = "-")))

p1

Perfect! very helpful. Is there a way to also force the new shading to the "back" of the plot so that the rectangles from the first plot are on top. — B. Davis, Sep 15 '15 at 17:52
Adding your new line of code to the initial call (eg: p1 <- ggplot(df, aes(x=IndID, fill = IndID))+ geom_rect(data = dateRanges, aes(ymin = start , ymax = end, xmin = -Inf, xmax = Inf), inherit.aes= F, alpha = 0.4, fill = c("lightblue"))+ geom_rect(aes(x = IndID, xmin = as.numeric(IndID) - 0.45, xmax = as.numeric(IndID) + 0.45, ymin = StartDate, ymax = EndDate))+... ) results in the error "Error: Discrete value supplied to continuous scale." I must be missing something else... — B. Davis, Sep 15 '15 at 18:01
Just realised that messed up your ordering, see the edit. Also cleaning up the comments — jeremycg, Sep 15 '15 at 20:18

Plotting rectangles in ggplot2 - Invalid input: time_trans works with objects of class POSIXct only

1 Answers1

Linked