Sharing axes only in some subplots of Plotly

Question

I have 3 subplots created with plot_ly (one for men, one for women and one for the gap). Since the gap is the subtraction of m-w, the ranges of m and w are far from the ranges of gap.

Is it possible to share axes only for m and w, and keep the axes for gap free?

shareY=TRUE and shareX=TRUE would make all three axes equal, but I only need to share two of them. Also, I don't want to plot the gap in a separate call of plot_ly because I need the legend to be linked to all the three plots (so I can hide/show markers).

Sample data:

library(dplyr)
library(plotly)
set.seed(197)
a <- tibble(
  pais = rep(c("ARGENTINA", "BOLIVIA", "BRASIL", "CHILE", "COLOMBIA"),each = 2),
  sexo = rep(c("m", "w"), 5),
  ea1 = rnorm(n=10, mean = 50, sd = 1),
  ea2 = rnorm(n=10, mean = 40, sd = 1))
a <- a %>% 
  group_by(pais) %>%
  summarise(sexo = "gap",
            across(ea1:ea2, ~ -diff(.x))) %>%
  bind_rows(a) %>% ungroup()

Plot:

a %>% group_by(sexo) %>%
  group_map(~ plot_ly(data=., x = ~ea1, y = ~ea2,
                      color = ~pais, type = "scatter", mode="markers",
                      showlegend = (.y == "m"),
                      legendgroup = ~pais), keep = TRUE) %>%
  subplot(nrows = 1, shareX = TRUE, shareY=TRUE)

Answer 1

You don't need to use group_map for the subplots to share a legend. If you use the same data to create the plots and legend, Plotly is going to gravitate towards keeping a single legend regardless. If you wanted to keep the gaps free, your best bet is nested subplots. For that, you'll need to pull them out of the group_map.

First, I'm going to create three separate plots: plt1, plt2, and plt3. Then I'm going to nest their subplots. I'm not entirely sure what you mean about having three plots but wanting the m/f to have a separate space. If they truly share x and y, there is only one plot.

# create a plot for each unique sexo
map(1:length(unique(a$sexo)),
    function(h) {
      df1 <- filter(a, sexo == unique(a$sexo)[h])
      p <- plot_ly(df1, x = ~ea1, y = ~ea2, color = ~pais,
                   type = "scatter", mode = "markers",
                   showlegend = (unique(a$sexo)[h] == "m"),
                   legendgroup = ~pais)
      assign(paste0("plt", h), p, envir = .GlobalEnv)
    })

Nest them.

subplot(plt1, subplot(plt2, plt3, shareY = T), widths = c(.33, .66))