Diagonal annotations in Plotly's categorical heatmap

Question

Consider the following R snippet to render a heatmap with categorical axes:

library(plotly)

x <- c("Blue", "Red")
y <- c("One", "Two")

plot_ly(x = x, y = y, z = c(1:4)) %>%
    add_heatmap(
        opacity = 0.9
    ) %>%
    add_annotations(
        text = c(1:4),
        showarrow = FALSE
    )

This renders the following heatmap:

The annotations appear to be distributed diagonally and unevenly, starting from the bottom left cell. 1 and 3 are in the bottom left cell, and 2 and 4 in the upper right. Why is this? How should my annotation text be structured for it to be ordered more intuitively (horizontally or vertically)?

Welcome to Stackoverflow! Your chances of getting an answer are much higher if you provide a [MCVE]. In your case, give us the values of `corrs`. — Maximilian Peters, May 09 '18 at 19:13
Hi Maximilian, and thanks for the welcome! Sorry about omitting corrs, I didn't think the contents were relevant when I posted. I've updated the OP to include the contents of the data frame. — Justin Sleep, May 09 '18 at 20:33
No worries! Here is a great thread for R question on Stackoverflow: https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example — Maximilian Peters, May 09 '18 at 20:53
Thanks, Max. I've updated the example in my question to be more minimal and easily reproduced. — Justin Sleep, May 10 '18 at 02:17

score 2 · Accepted Answer · answered May 10 '18 at 08:36

I can only speculate about the problem but in the provided image you can see that Plotly only used two values out of the 4 z-values. Your colorscale on the right goes from 1 to 2, not 1 to 4. This happens mMost likely because you provided only two x and y values.

Use a data frame

df <- expand.grid(x, y)
df <- transform(df, text = paste(Var1, Var2, sep='_'))

print(df)

Var1 Var2     text
1 Blue  One Blue_One
2  Red  One  Red_One
3 Blue  Two Blue_Two
4  Red  Two  Red_Two

You can now easily use add_annotations

add_annotations(x = df$Var1,
                y = df$Var2,
                text = df$text)

To get the following plot

Complete code

library(plotly)

x <- c("Blue", "Red")
y <- c("One", "Two")
df <- expand.grid(x, y)
df <- transform(df, text = paste(Var1, Var2, sep='_'))

p <- plot_ly(x = df$Var1, 
             y = df$Var2, 
             z = c(1:4)
             ) %>%
  add_heatmap(opacity = 0.9
              ) %>% 
  add_annotations(x = df$Var1,
                  y = df$Var2,
                  text = df$text)

p

Alternatively you could loop over your values and add an annotation for each one.

library(plotly)

x <- c("Blue", "Red")
y <- c("One", "Two")

p <- plot_ly(x = x, 
             y = y, 
             z = c(1:4)
             ) %>%
  add_heatmap(opacity = 0.9)

for (val_x in x)
{
  for (val_y in y)
  {
    p <- add_annotations(p, 
                         x = val_x,
                         y = val_y,
                         text = paste(val_x, val_y, sep = '_'))
  }
}

p

Thanks, Maximilian! Your answer pointed me in the right direction. I ended up using melt() from the reshape library to create the annotations to pass in. — Justin Sleep, May 10 '18 at 12:08

Diagonal annotations in Plotly's categorical heatmap

1 Answers1