How to choose variable to display in tooltip when using ggplotly?

Question

I have a simple data frame:

seq <- 1:10
name <- c(paste0("company",1:10))
value <- c(250,125,50,40,40,30,20,20,10,10)
d <- data.frame(seq,name,value)

And I want to plot it this way:

require(ggplot2)
ggplot(data = d,aes(x=seq,y=value))+geom_line() + geom_point()

Now I want to use plotly, mostly to be able, when mousing over a point, to get other information than the value, such as the company name. I try this :

require(plotly)
ggplotly()

which get me a tooltip, but with only seq and value. I tried the option tooltip= but it's specified you can use the only variable describe in the aesthetic, and I don't use the name in my aes.

Any solution? I saw I am not the first with this problem, but I haven't found answer working with ggplotly.

Solution 1

You don't need to modify the plotly object as suggested by @royr2. Just add label = name as third aesthetic

ggplot(data = d, aes(x = seq, y = value, label = name)) + geom_line() + geom_point()

and the tooltip will display name in addition to seq and value.

The ggplotly help file says about tooltip parameter:

The default, "all", means show all the aesthetic mappings (including the unofficial "text" aesthetic).

So you can use the label aesthetic as long as you don't want to use it for geom_text.

BTW: I've also tried text instead of label

ggplot(data = d, aes(x = seq, y = value, text = name)) + geom_line() + geom_point()

but then ggplot2 complained

geom_path: Each group consists of only one observation. Do you need to adjust the group aesthetic?

and plotted only points. I had to add a dummy group to geom_line to remove the issue:

ggplot(data = d, aes(x = seq, y = value, text = name)) + geom_line(group = 1) + geom_point()

(But note if you put the dummy group as fourth aesthetic inside aes() it will appear by default also in the tooltip.)

However, I find the unofficial text aesthetic can become useful alongside label if you want to have different strings plotted by geom_text and shown in the tooltip.

Edit to answer a question in comments: The tooltip parameter to ggplotly() can be used to control the appearance. ggplotly(tooltip = NULL) will suppress tooltips at all. ggplotly(tooltip = c("label")) selects the aesthetics to include in the tooltip.

Solution 2

Building on @UweBlock's answer, you can also create dummy aesthetics in order to display multiple labels in tooltips. I can't find where this is documented, but discovered it emperically. The dummy variables show up in the order you specify them, but priority is given to the default variables (e.g. x and y). To get around this, you can specify those variables in a separate aesthetic, as shown below:

library(plotly)
p = ggplot(iris, aes(label=Species, label2=Petal.Length, label3=Petal.Width)) + 
  geom_point(aes(Sepal.Length,Sepal.Width))
ggplotly(p)

Solution 3

The unofficial text aesthetic allows you to introduce all the variables you want (here I use name twice to show it):

require(ggplot2)
ggplot(data = d,aes(x = seq, 
                    y = value,
                    group = 1,
                    text = paste('name: ', name,
                                 '</br>name_again: ', name)
                    ))+
  geom_line() + 
  geom_point()

I have to add a dummy group aesthetic for the geom_line to work properly as @UweBlock suggested.

At last I choose what I want to show in the tooltip (here I excluded group).

require(plotly)
ggplotly(, tooltip = c("x", "y", "text"))

Solution 4

You'll have to modify the plotly object to do this. Or use plot_ly() to create the graph instead...

EDIT: With the release of plotly 4.0 the syntax will change a little bit.

seq <- 1:10
name <- c(paste0("company",1:10))
value <- c(250,125,50,40,40,30,20,20,10,10)
d <- data.frame(seq,name,value)

require(plotly)
gg <- ggplot(data = d,aes(x=seq,y=value))+geom_line() + geom_point()

gg <- plotly_build(gg)

#OLD:
gg$data[[1]]$text <- paste("Seq:", d$seq, "<br>",
                           "Value:", d$value, "<br>",
                           "Company:", d$name)

#UPDATED: 
#Plotly_build creates two separate traces:
#One with mode = markers and the other with mode = lines
#Hence modify text for the second trace

gg$x$data[[2]]$text <- paste("Seq:", d$seq, "<br>",
                           "Value:", d$value, "<br>",
                           "Company:", d$name)

gg

View more solutions

Comments

[Malta]

great, it works thank you ! I will give plot_ly() a look too, I imagine this can give more readable code, but this way I am not compulsed to update all my code !

[Malta]

thank you, this seems to me to be a more readable way of obtaining what I want. If someone want to use geom_text, then aes should be put directly in geom_line and in geom_text.

[Joy]

This solution works for me locally but when I place the app on shiny server the additional tooltip is not there. Then I tried the solution below and that works.

[Joy]

This really works. I have a quick question, is there any way to order the tooltip? I mean x and y axis will be first and then the other labels?

[teadotjay]

@Joy it seems to want to put x and y first, unless you move them to the geom_point's aes: p = ggplot(iris, aes(label=Species, label2=Petal.Length, label3=Petal.Width)) + geom_point(aes(Sepal.Length,Sepal.Width)). The resulting plot will show the tooltips in the order they appear in the command.

[Joy]

Thank you for your response but I am getting error: argument 3 is not a vector. Any idea?

[teadotjay]

Not sure; I pasted that line into R (version 3.2.5) and got the expected plot. I'm using plotly version 3.4.13 and ggplot2 version 2.1.0. See my updated answer for the full script.

[Cyrus Mohammadian]

@royr2 still no dice -"company" does not appear in tooltip

[royr2]

@CyrusMohammadian seems okay for me - I am using plotly v4.0.1

[Cyrus Mohammadian]

@royr2 interesting, I'm on version 3.6.0, I just reinstalled it from CRAN and it's still that same version, are you installing from github?

[Cyrus Mohammadian]

I've updated plotly by installing it from github, it works now for the example above but when used with a ggplot choropleth map (geom_polygons), it adds an additional tooltip box and keeps the previous one as well...very odd behavior

[mattek]

How to make it work with facets present? It works only in the 1st facet for me.

[Jan Stanstrup]

you need to run through all of them. So you can make a for loop (lets say with i in....) where you change gg$data[[i]]$text. Seems in plotly 4.5.6.9000 you need to change gg$x$data instead of gg$data.

[Jack H]

Is it possible to also remove the default labels in addition to adding others?

[Uwe]

@VisionIncision Yes. I've updated my answer accordingly.

[Jack H]

@UweBlock Thank you for your assistance thus far. Sorry I should have been clearer, for some reason with the way I am creating a plot (possibly because I am doing it with shiny), it does not seem to make a difference if I use any tooltips arguments. So I was wondering if there was something I could do in the aes() function, like how you can add them with label=, label2= etc to remove default labels.

[Nova]

@JackH I struggled with this for a while. Remember to use the name of the aesthetic, not your variable, in the call to tooltip. For example, use "x" not "seq" and "label" not "name".

[Rosemary]

Introducing the "text" in aes() really solved my problem. THANK YOU!!

[stats_noob]

Can you please take a look at this question? stackoverflow.com/questions/66166106/… thanks!

[Nate]

this is still working in R v4.0.3. I really wish I could up vote this post twice because it's an awesome trick

[mal]

This feels like black magic but it works!

[hugh-allan]

I found that if you use label = name the tooltip info will show name: companyX, whereas if you use text = name the tooltip will just show companyX. This was useful for me because I had already created the tooltip info in it's own column using str_c('Line 1: ', line1_col, '<br> Line 2: ', line2_col', ...) and when using label = it annoyingly included the name of the column at the beginning of the label.