I would like to add trace of a median line on my box plot.
like this
Here are my plots so far:
library(plotly)
p <- plot_ly(y = ~rnorm(50), type = "box") %>%
add_trace(y = ~rnorm(50, 1))
p
-
Do you mind to produce a full mcve? I can achieve this pretty easily with python and usually the the two versions have the same functionalities.rpanai– rpanai2020-02-06 22:03:13 +00:00Commented Feb 6, 2020 at 22:03
-
@Sup'A How did my suggestion work out for you?vestland– vestland2020-03-29 21:58:14 +00:00Commented Mar 29, 2020 at 21:58
Add a comment
|
2 Answers
Just start out with a scatter plot using plot_ly(..., type='scatter', mode='lines', ...), and follow up with one add_boxplot(...' inherit=FALSE, ...) per box plot. Here's how you do it for an entire data.frame:
Complete code with sample data:
library(dplyr)
library(plotly)
# data
df <- data.frame(iris) %>% select(-c('Species'))
medians <- apply(df,2,median)
# create common x-axis values for median line and boxplots
xVals <- seq(0, length(medians)-1, by=1)
# plotly median line setup
p <- plot_ly(x = xVals, y=medians, type='scatter', mode='lines', name='medians')
# add a trace per box plot
i <- 0
for(col in names(df)){
p <- p %>% add_boxplot(y = df[[col]], inherit = FALSE, name = col)
i <- i + 1
}
# manage layout
p <- p %>% layout(xaxis = list(range = c(min(xVals)-1, max(xVals)+1)))
p
Another option is to use ggplot2 and convert it into plotly
library(ggplot2)
library(dplyr)
library(tidyr)
library(plotly)
p = iris %>% pivot_longer(-Species) %>%
ggplot(aes(x=name,y=value,col=name)) +
geom_boxplot() + stat_summary(inherit.aes = FALSE,
aes(x=name,y=value,group=1),fun.y=median,geom="line")
ggplotly(p)
A brief explanation of the code, I use pivot_longer from tidyr to cast the data frame into long format, and first made the boxplot with the column names as x variable and color.
The stat_summary part, I specified again the same x and y variables again, and omitted the colour this time, adding group=1, this tells stat_summary to consider the whole data frame as one group, and to summarize all the y values per x-group, and draw a line through it.
