Expand y axis to negative numbers on a log10 scale axis in ggplot2

Question

I am plotting a combination of points and lines with 2 sets of labels. I want to plot the labels above and below the points as you can see below. At the same time to make it better to read I have used the log10 scale on the y axis and everything is fine except some of the labels are outside of the chart area and I have used and tried every method suggested in many posts to see if I get any favorable result or not. I am looking for either of the solutions:

1- expand the y axis to negative so that the labels can be seen. Note that ylim or limits=c(x,y) is not working for log scale or sqrt scale if the numbers are negative

2- trick the geom_text to make the labels be visible regardless of the y limits. Note that I have tried vjust="inward" and it is working ok, but then I have to use geom_text_repel which moves the labels around and makes it hard to read, so I still like to place the labels right on top and bottom of the points

any helps is appreciated!

Here is the code to generate the data frame:

df1_InSAP_Only <- structure(list(Year_Month = c(
    "2016_06", "2016_06", "2016_07", 
    "2016_07", "2016_08", "2016_08", "2016_09", "2016_09", "2016_09", 
    "2016_09", "2016_10", "2016_10", "2016_10", "2016_10", "2016_11", 
    "2016_11", "2016_12", "2016_12", "2017_01", "2017_01", "2017_01", 
    "2017_02", "2017_02", "2017_02", "2017_02", "2017_03", "2017_03", 
    "2017_03", "2017_03", "2017_03", "2017_03", "2017_04", "2017_04", 
    "2017_04", "2017_04", "2017_04", "2017_05", "2017_05", "2017_05", 
    "2017_05", "2017_05", "2017_05", "2017_06", "2017_06", "2017_06", 
    "2017_06", "2017_06", "2017_06", "2017_07", "2017_07"), 
    Business = c("A", 
     "E", "A", "B", "B", "E", "F", "A", "H", "B", "A", "D", "B", "E", 
     "B", "E", "F", "B", "F", "B", "E", "A", "B", "C", "E", "F", "A", 
     "G", "D", "B", "E", "F", "A", "G", "B", "E", "F", "A", "D", "B", 
     "C", "E", "F", "A", "D", "B", "C", "E", "F", "A"),
    `MMR Count` = c(2L, 
      1L, 1L, 7L, 2L, 1L, 1L, 3L, 1L, 5L, 1L, 1L, 4L, 1L, 8L, 4L, 1L, 
      4L, 2L, 2L, 2L, 3L, 8L, 1L, 2L, 1L, 7L, 1L, 4L, 9L, 2L, 4L, 10L, 
      2L, 15L, 7L, 4L, 27L, 2L, 14L, 1L, 6L, 9L, 31L, 5L, 14L, 1L, 
      4L, 5L, 21L),
     `Duration Average` = c(37, 20, 9, 8, 2, 5, 1, 1, 
      1, 14, 1, 19, 8, 1, 21, 77, 1, 18, 8, 1, 1, 194, 9, 14, 19, 1, 
      10, 1, 6, 9, 18, 4, 12, 170, 7, 35, 9, 10, 7, 12, 3, 15, 5, 9, 
      10, 10, 18, 11, 16, 14)), .Names = c("Year_Month", "Business", 
      "MMR Count", "Duration Average"), row.names = c(NA, 50L), class = "data.frame")

Here is the code that generates the plot:

library(ggplot2)
ggplot(df1_InSAP_Only,
        aes(x=Year_Month,
            y=`Duration Average`,
            group=Business,
            color=Business,
            size=`MMR Count`)) +
  geom_line(aes(group=Business),stat="identity", size=1, alpha=0.7) +
  geom_point(aes(colour=Business, alpha=0.7)) +
  facet_wrap(~ Business, ncol=2) +
  scale_y_log10( limits=c(-100,1000),breaks=c(0,1,10,100,1000)) +   
  scale_alpha_continuous(range = c(0.5,1), guide='none') + #remove the legend for alpha
  geom_text(data=. %>% dplyr::group_by(Business),
            aes(label=`Duration Average`,vjust=-2),
            size=3,
            position = position_dodge(width=0.9)) +
  geom_text(data=. %>% dplyr::group_by(Business),
            aes(label=`MMR Count`,vjust=3),
            size=3,
            position = position_dodge(width=0.9),
            color="brown")

and here is the plot:

*"Unfortunately, I cannot give you a reproducible dataset because the whole data is pulled from a database and undergo many calculations"* We don't need your real data - we just need *something* with similar structure that illustrates the problem - something to to demonstrate a solution on. 10-20 rows if plenty. Popular choices are built-in data and simulated data. — Gregor Thomas, Jul 13 '18 at 17:55
As a side-note, you should never use `data$column` inside `aes()`. Change `group=df1_InSAP_Only$Business` to `group = Business`. And I can't imagine that `dplyr::group_by(Business)` is doing anything useful inside the plot layer. — Gregor Thomas, Jul 13 '18 at 17:58
@Gregor `you should never use data$column inside aes()`, why? — Ibo, Jul 13 '18 at 18:03
it's not how ggplot2 was create to work (not the most optimal way to use it), please go through `ggplot2` tutorials for that — pogibas, Jul 13 '18 at 18:04
[R - ggplot2 - difference between ggplot(data, aes(x=variable…)) and ggplot(data, aes(x=data$variable…))](https://stackoverflow.com/a/51194689/903061) — Gregor Thomas, Jul 13 '18 at 18:24

Gregor Thomas · Accepted Answer · 2018-07-13T18:45:01.493

You can't make a logged y-scale go negative - logs of a negative number are undefined. Just make it go closer to 0. Here's your graph with

scale_y_log10(limits=c(.1, 1000),breaks=c(1, 10, 100, 1000))

If you want more (will depend on the size of the final plot, size of the text, amount of your vjust), go to 0.05, or 0.01...

I'd also highly recommend using a Date format for your x-axis data, look how much nicer these axis labels are (and how the plot looks cleaner with fewer vertical gridlines).

df1_InSAP_Only$date = as.Date(paste0(df1_InSAP_Only$Year_Month, "_01"), format = "%Y_%m_%d")

 # use date column on x-axis
 # reduce vjust amounts
 # get rid of meaningless group_by() statements
 # get rid of unused position dodges
ggplot(df1_InSAP_Only,
        aes(x=date,
            y=`Duration Average`,
            group=Business,
            color=Business,
            size=`MMR Count`)) +
  geom_line(aes(group=Business),stat="identity", size=1, alpha=0.7) +
  geom_point(aes(colour=Business, alpha=0.7)) +
  facet_wrap(~ Business, ncol=2) +
  scale_y_log10( limits=c(.1,1000),breaks=c(1,10,100,1000)) +   
  scale_alpha_continuous(range = c(0.5,1), guide='none') + #remove the legend for alpha
  geom_text(aes(label=`Duration Average`,vjust=-1),
            size=3) +
  geom_text(aes(label=`MMR Count`,vjust=2),
            size=3,
            color="brown")

I am trying to add the mean of data as a horizontal line, but I could not get around it, I posted a new question for it: https://stackoverflow.com/questions/51332140/how-to-add-the-mean-of-data-as-a-horizontal-line-on-a-faceted-plot-in-r-ggplot2 — Ibo, Jul 13 '18 at 20:05

Expand y axis to negative numbers on a log10 scale axis in ggplot2

1 Answers1