CODEWITHSUNDEEP
rggplot2plot
st4co4
st4co4

Reputation: 467

Plotting parallel coordinates with confidence intervals in R

Is there an R package allowing to plot parallel coordinates with confidence intervals? Like the example below.

Or what would be the easiest alternative? Making multiple ggplots and printing them side by side with ggarrange or patchwork?

enter image description here

Generated data

library(tidyverse)

set.seed(1)

#make new data
df = tibble(
  factor= c("A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M", "N", "O"),
  var1 = rnorm(15, 50, 20),
  var2 = rnorm(15, 10, 4),
  var3 = rnorm(15, 2, 1))

#add confidence intervals
df = df %>% mutate(
  var1_lower = var1 - runif(1, 5, 15),
  var1_upper = var1 + runif(1, 5, 15),
  var2_lower = var2 - runif(1, 0.5, 2.2),
  var2_upper = var2 +  runif(1, 0.5, 2.2),
  var3_lower = var3 - runif(1, 0.1, 0.6),
  var3_upper = var3 + runif(1, 0.1, 0.6))

#reorder columns
df = df[c("factor", "var1", "var1_lower", "var1_upper", "var2", "var2_lower", "var2_upper", "var3", "var3_lower", "var3_upper")]

enter image description here

Upvotes: 0

Views: 297

Answers (1)

Gopal Srivastava
Gopal Srivastava

Reputation: 155

There are two libraries that let you plot parallel coordinate charts.

  1. GGally
  2. MASS
  3. ggparci

I believe ggparci would be best fit for you question. As it allows you to plot confidence interval with the data. I have given the link to the library documentation in the comment above.

But, I am mentioning rest two for the benefit of everyone.

In GGally library you can use

ggparcoord(df, columns=1:10, groupColum = 5)

GGally has scale option for further processing of the plot.

I plotted df without any processing and it looks like : enter image description here

And in MASS library there is a function called parcoord()

Example use:

# MASS library
library(MASS)

# Color
color <- colors()[as.numeric(iris$Species)*11]

# Plot
parcoord(iris[,c(1:4)] , col= color)

Upvotes: 1

Related Questions