R | 繪製箱線圖

箱線圖是能同時反映數據統計量和整體分佈展示圖，Boxplot從下到上展示的是最小值，第一四分位數 (箱子的下邊線)、中位數 (箱子中間的線)、第三四分位數 (箱子上邊線)、最大值。
數據來源 ：CRA_jobs.txt（通過爬蟲獲得獵聘網職位CRA酬薪）

library(ggplot2)
library(ggthemes)
library("RColorBrewer")
library(dplyr)
library(tidyr)
library(magrittr)
display.brewer.all()  # 顯示所有可用色板
display.brewer.all(type = "seq")  # 查看漸變色板
###顏色設置
cols <- brewer.pal(12,"Set3")

###導入數據
CRA_jobs <- read.csv("CRA_jobs.txt", header = T, sep = "\t", stringsAsFactor=F)

### 篩選數據
major_city <- c("上海", "北京", "深圳", "廣州", "福州", "廈門")
CRA_major_city <- filter(CRA_jobs, city %in% major_city )
CRA_major_city$city <- factor(CRA_major_city$city, levels = c("上海", "北京", "深圳", "廣州", "福州", "廈門"), labels = c("上海", "北京", "深圳", "廣州", "福州", "廈門"))
CRA_major_city <- filter(CRA_major_city, salary <=50)

### 可視化 
plot <- ggplot(data = CRA_major_city,  aes(x=city, y=salary)) +
  stat_boxplot(geom ='errorbar', width=0.3, position=position_dodge(1.05)) +
  geom_boxplot(aes(fill =factor(city))) +
  scale_x_discrete(name = "城市") +
  scale_y_continuous(name = "薪水") +
  scale_fill_manual(values = cols) +
  guides(fill = guide_legend(title = "城市", title.position = "top", # 指定圖例名稱爲"城市", 位置爲箱體的上邊
                             key.width = 3, key.height = 10, nrow = 3, ncol = 2, byrow = TRUE)) + # 修改箱體尺寸，並矩形排列，按行排
  theme_bw()    ### 設置主題，直接應用theme包
# theme(title = element_text(family = "sans", face = "bold"), panel.grid.major = element_blank()) .... ### 自定主題
ggsave("CRA_major_city.png", plot, width = 6.15, height = 3.15)

### 分組可視化，整理數據
## 方法一
CRA_major_city$group <- CRA_major_city$city %>% lapply(function(x){
  x <- ifelse(as.character(x) %in% c("上海", "北京"), "G1",ifelse(as.character(x) %in% c( "深圳", "廣州"), "G2",ifelse(as.character(x) %in% c("福州", "廈門"), "G3",0)))
}) %>% unlist() %>% as.factor()
## 方法二'
groups <- data.frame(areas=c("上海","北京","深圳","廣州","福州","廈門"), group=c("G1","G1","G2","G2","G3","G3"))
CRA_major_city$group <- groups[CRA_major_city$city,]$group

ggplot() +
  #stat_boxplot( geom ='errorbar', width=0.3, position=position_dodge(1.05)) +
  geom_boxplot(data = CRA_major_city,  aes(x=group, y=salary, fill =factor(city)), position=position_dodge(1.05)) +
  scale_x_discrete(name = "城市") +
  scale_y_continuous(name = "薪水") +
  scale_fill_manual(values = cols) +
  guides(fill = guide_legend(title = "城市", title.position = "top", # 指定圖例名稱爲"城市", 位置爲箱體的上邊
                             key.width = 3, key.height = 10, nrow = 3, ncol = 2, byrow = TRUE)) + # 修改箱體尺寸，並矩形排列，按行排
  theme_bw()    ### 設置主題，直接應用theme包