custom colors in R Plotly

I came across this question today because I was trying to define colors for every individual marker and struggling. The answer came from one of the other answers, but that answer said it didn't work for this question, but it did for mine.

Here is a gist with and MRE: https://gist.github.com/bhive01/d4821f6d73ea0f428767d9bb68fdd404

Essentially, rather than define colors with color and/or colors, you use marker = list(color = ~color). This assumes you have data like mine where every point has a color. This was critical for me because when I used color = ~Hex, colors = ~Hex the colors were used, but the data got sorted and the colors did not match their correct marker.

You need to specify a factor for the color parameters, and then a vector of colours for the colors parameter.

Here is a simple solution. Note the ordering required on the data frame before plotting.

require(dplyr)
require(plotly)

set.seed(42)

df <- data.frame(x = rep(LETTERS[1:5], 3), 
                 y = rexp(15, rate = 0.5),
                 z = c(rep("Adam", 5), rep("Arthur", 5), rep("Ford", 5)))
df <- arrange(df, desc(z))

plot_ly(df, 
        x = x, 
        y = y, 
        color = z, 
        colors = c("grey50", "blue", "red"), 
        type = "bar") %>% 
    layout(barmode = "stack")

The ordering on the data frame matters strangely. I would have thought plot_ly would use the order of the levels but it doesn't.

EDIT: This example uses plotly 3.x.x. If you use plotly 4.x.x or above, this code may not work as is. See here for more details: https://www.r-bloggers.com/upgrading-to-plotly-4-0-and-above/

Here's something that might help. Note that for some reason, Method 1 causes the legend entries to be black. So I am also suggesting a workaround.

library(plotly)
library(dplyr)
library(data.table)

mtcars$color <- factor(mtcars$gear, labels = c("blue", "red", "green"))

# Method 1
# Legend entries are all black
plot_ly(mtcars, x = as.factor(cyl), y = mpg, group = gear, type = "bar", marker = list(color = color), name = "test") %>% 
  layout(barmode = "stack", showlegend = T)

# Method 2
# Workaround
dt <- data.table(mtcars)

p <- dt[gear == 3,] %>% 
  plot_ly(x = factor(cyl), y = mpg, name = "Gear = 3", type = "bar", marker = list(color = "blue"))

p <- dt[gear == 4,] %>% 
  add_trace(x = factor(cyl), y = mpg, name = "Gear = 4", type = "bar", marker = list(color = "red"))

p <- dt[gear == 5,] %>% 
  add_trace(x = factor(cyl), y = mpg, name = "Gear = 5", type = "bar", marker = list(color = "green"))

p <- layout(p, barmode = "stack")

p