Setting margin of multiple pie charts using plotly in R

Question

I'm trying to set the margins between the pie charts, but now margin doesn't work and the pictures overlap each other.

labels <- rep(LETTERS[seq(from = 1, to = 11)],12)
values <- runif(length(labels), min=0, max=1)
group <- letters[seq(from = 1, to = 12)]
row=rep(seq(0,3), each=3)
col=rep(seq(0,2), time=4)
d1 <- data.frame(labels, values, group)

library(plotly)
p <- plot_ly() 
for (i in 1:12) {
  var = group[i] 
  d <- d1 %>% filter(group==var) 
  p <- p %>%  add_pie(data=d, labels = ~labels, values = ~values, type = 'pie',
                      textposition = "outside", textinfo = 'label+percent',
                      marker = list(colors = ~colors),
                      sort = F,
                      domain = list(row = row[i], column = col[i]),
                      margin = list(b=180, l=180, r=180, t=180) **#doesn't work**
  )
}

p <- p %>% layout(title = NULL, showlegend = F,
                  grid=list(rows=max(row)+1, columns=max(col)+1),
                  margin = list(b=80, l=80, r=80, t=80), **#doesn't work**
                  xaxis = list(showgrid = FALSE, zeroline = FALSE, showticklabels = FALSE),
                  yaxis = list(showgrid = FALSE, zeroline = FALSE, showticklabels = FALSE))

p

orca(p, "pie_chart.pdf", scale = 1, width = 1200, height = 1600)

I've also tried setting up to use the subplot function, but none of that works. Also I've tried drawing with ggplot2 but can't set up labels like plotly. I would like to use plotly to draw the labels without overlapping or ggplot2 to draw the labels like the above illustration. Thank you in advance for answers.

Answer 1

I see that you used domain to set the row and column, but if you want a buffer between rows and columns, use domain to set the real estate in 'plot space'. By default the range is 0 to 1.

So for horizontal placement, plot space plot space plot - or 3/10 + 1/2 + 3/10 + 1/2 + 3/10. You can, of course, use different spacing - the goal is that whatever sizes you use, that the total amounts to 1.

For vertical placement, there's 4 plots and 3 spaces. Here I used .225 for each plot and .1/3 (1/3 of .1 remaining after .225 * 4 = 0.9).

One other element I changed was the font size of the labels. In layout, I added font = list(size = 6)

I create 3 new constants.

buff = .05; rw = 4; cl = 3

Then I updated the documentation of the domains in your for loop using discrete math.

for (i in 1:12) {
  var = group[i] 
  d <- d1 %>% filter(group == var) 
  p <- p %>%  add_pie(
    data = d, labels = ~labels, values = ~values, type = 'pie',
    textposition = "outside", textinfo = 'label+percent',
    marker = list(colors = ~colors),
    sort = F,
    domain = list( # if it's a plot on the edges or not
      x = c(ifelse(col[i] == 0, 0, col[i] * 3/10 + buff * col[i]),    # horizontal
            ifelse(col[i] == 2, 1, (col[i] + 1) * 3/10 + buff * col[i])),  
      y = c(ifelse(row[i] == 0, 0, (1/rw - buff/2 + .1/3) * row[i]),  # vertical 
            ifelse(row[i] == 3, 1, (1/rw - buff/2) * (row[i] + 1) + (.1/3 * row[i]))))
  )
}

This equates to the following domains for the four plots (if you were to manually write it).

x0 = 0    x1 = 0.3  y0 = 0      y1 = 0.225
x0 = 0.35 x1 = 0.65 y0 = 0      y1 = 0.225
x0 = 0.7  x1 = 1    y0 = 0      y1 = 0.225
x0 = 0    x1 = 0.3  y0 = 0.2583 y1 = 0.483
x0 = 0.35 x1 = 0.65 y0 = 0.2583 y1 = 0.483
x0 = 0.7  x1 = 1    y0 = 0.2583 y1 = 0.483
x0 = 0    x1 = 0.3  y0 = 0.516  y1 = 0.7416
x0 = 0.35 x1 = 0.65 y0 = 0.516  y1 = 0.7416
x0 = 0.7  x1 = 1    y0 = 0.516  y1 = 0.7416
x0 = 0    x1 = 0.3  y0 = 0.775  y1 = 1
x0 = 0.35 x1 = 0.65 y0 = 0.775  y1 = 1
x0 = 0.7  x1 = 1    y0 = 0.775  y1 = 1

With the updates I've made, here's the plot.

If you were to change the vertical alignment to .2 for plots and .2/3 for spacing, this is what you would get. (Change the buff/2 to buff and the.1/3 to .2/3 for y domain)