多个箱形图并排 [英] Multiple boxplots side by side

问题描述

我正在尝试与ggplot2并排制作多个箱形图.我一直在关注stes

---

在ggplot2中，涉及更多工作:

  set.seed(11)数据<-data.frame(原始= rnorm(1000)，正常= rnorm(1000)，学生= cbind(c(rt(1000，df = 2))，c(rt(1000，df = 3))))图书馆(dplyr)图书馆(tidyr)库(ggplot2)data2< -pivot_longer(数据，cols = everything())％>％mutate(name = factor(name，level = c("Raw"，"Normal"，"Student.1"，"Student.2"))))data3<-data2％>％摘要(min = min(value)，max = max(value))p1<-filter(data2，name％in％c("Raw"，"Normal"))％&％ggplot(aes(y = value，fill = name))+geom_boxplot()+facet_grid(〜name)+ylab("Log-returns")+ylim(data3 $ min，data3 $ max)+theme_bw()+主题(legend.position =无"，axis.ticks.x = element_blank()，panel.grid.major.x = element_blank()，panel.grid.minor.x = element_blank()，axis.text.x = element_text(color ="white"))p2<-filter(data2，grepl("Student"，name))％&％mutate(what ="Student")％>％ggplot(aes(x = name，y = value，fill = name))+geom_boxplot()+scale_fill_manual(values = c("green1"，"green3"))+scale_x_discrete(labels = c("df = 2"，"df = 3"))+facet_grid(〜什么)+ylim(data3 $ min，data3 $ max)+theme_bw()+主题(legend.position =无"，axis.title.y = element_blank()，axis.title.x = element_blank()，axis.text.y = element_blank()，axis.ticks.y = element_blank())图书馆(ggpubr)ggarrange(p1，p2) 

---

I'm trying to make multiple boxplots with ggplot2 side by side. I've been following the stes Multiple boxplots placed side by side for different column values in ggplot but without much luck.

I have the following dataframes

Raw <- sp500_logreturns
Normal <- rnorm(1000, 0, sd(sp500_logreturns)
Student <- cbind(c(rt(1000, df = 2)),c(rt(1000, df = 3)))

And I want to make the following Boxplot sketch

My Raw vector contains logreturns-transformation of my prices downloaded as an environment from yahoo into R. I must admit I'm quite lost, and do not know if I'm on an impossible mission. I hope I've described my problem well enough together with my sketch. Thank you in advance.

Update 1: The goal is to compare the raw data distribution (which is leptokurtic) and therefore a student disitribution with 2 or 3 degree of freedom might be more suitable than a normal distribution. To give you an idea of the data I'm looking at, here's a summary:

      Min.    1st Qu.     Median       Mean    3rd Qu.       Max. 
-0.0418425 -0.0023740  0.0005898  0.0004704  0.0045065  0.0484032  

Here is my boxplot made from Edward's code: Boxplot (Edward)

Update 2: I figured it out. I used fitdist from rugarch to find out the best student distribution fitted to the raw data. This way I could ignore trying to match different dfs of the student distribution. This is what I will go on with:

fitdist(distribution = 'std', sp500_logreturns)$pars
          mu        sigma        shape 
0.0008121004 0.0113748869 2.3848231857 

data <- data.frame(
        Raw = as.numeric(sp500_logreturns),
        Normal = rnorm(1006, 0, sd(sp500_logreturns)),
        Student = rdist(distribution = 'std', n = 1006, mu = 0.0008121004, sigma = 0.0113748869, shape = 2.3848231857)
)

data2 <- pivot_longer(data, cols=everything()) %>%
        mutate(name=factor(name, levels=c("Raw","Normal","Student")))

data3 <- data2 %>% summarise(min=min(value), max=max(value))

pbox1 <- (filter(data2, name %in% c("Raw","Normal","Student")) %>%
        ggplot(aes(y=value, fill=name)) +
        geom_boxplot() +
        facet_grid(~name) +
        ylab("Log-returns") +
        ylim(data3$min, data3$max) +
        theme(legend.position = "none",
              axis.ticks.x=element_blank(),
              panel.grid.major.x = element_blank(),
              panel.grid.minor.x = element_blank(),
              axis.text.x=element_text(color="white"))+
        ggtitle("Boxplot comparison")+
        theme(plot.title = element_text(hjust = 0.5)))

And this gives me: Boxplot (final)

解决方案

In base R:

set.seed(11)
data <- data.frame(
  Raw = rnorm(1000),
  Normal = rnorm(1000),
  Student = cbind(c(rt(1000, df = 2)),c(rt(1000, df = 3)))
)

ylim=c(min(data), max(data))

layout(matrix(1:3, nc=3), widths=c(5,4,5))
par(las=1, mar=c(2,4,5,0))
boxplot(daat$Raw, col="steelblue", ylab="Log-returns", ylim=ylim)
title(main="Raw", line=1)

par(mar=c(2,1,5,0))
boxplot(data$Normal, yaxt="n", col="tomato", ylim=ylim)
title(main="Normal", line=1)

par(mar=c(2,1,5,1))
boxplot(data[,3:4], yaxt="n", col=c("green1","green3"), names=c("df = 2","df = 3"), ylim=ylim)
title(main="Student", line=1)
title(main="Boxplot comparison", outer=TRUE, line=-1.5, cex.main=1.5)

---

---

In ggplot2, more work is invovled:

set.seed(11)
data <- data.frame(
  Raw = rnorm(1000),
  Normal = rnorm(1000),
  Student = cbind(c(rt(1000, df = 2)),c(rt(1000, df = 3)))
)

library(dplyr)
library(tidyr)
library(ggplot2)

data2 <- pivot_longer(data, cols=everything()) %>%
  mutate(name=factor(name, levels=c("Raw","Normal","Student.1","Student.2")))

data3 <- data2 %>% summarise(min=min(value), max=max(value))

p1 <- filter(data2, name %in% c("Raw","Normal")) %>%
  ggplot(aes(y=value, fill=name)) +
  geom_boxplot() +
  facet_grid(~name) +
  ylab("Log-returns") +
  ylim(data3$min, data3$max) +
  theme_bw() +
  theme(legend.position = "none",
        axis.ticks.x=element_blank(),
        panel.grid.major.x = element_blank(),
        panel.grid.minor.x = element_blank(),
        axis.text.x=element_text(color="white"))

p2 <- filter(data2, grepl("Student", name)) %>%
  mutate(what="Student") %>%
  ggplot(aes(x=name, y=value, fill=name)) +
  geom_boxplot() +
  scale_fill_manual(values=c("green1","green3")) +
  scale_x_discrete(labels=c("df=2", "df=3")) +
  facet_grid(~what) +
  ylim(data3$min, data3$max) +
  theme_bw() +
  theme(legend.position = "none",
        axis.title.y = element_blank(),
        axis.title.x=element_blank(),
        axis.text.y=element_blank(),
        axis.ticks.y=element_blank())

library(ggpubr)
ggarrange(p1, p2)

---

这篇关于多个箱形图并排的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！