使用R对Sankey/Alluvial图中的部分变量进行美化排序

Question

我正在努力提高我在数据可视化方面的技能，我几乎得到了我想要的。但在某些时候，我被卡住了，无法再前进了。请注意，伙计们，我在这里做了广泛的研究，试图找出我的疑惑，这对我有很大帮助。

这是我的数据集：

https://app.box.com/s/pp5p5chgypn6ba33anotie7wlxvdu01v

这是我的代码：

library(tidyverse)
library(ggalluvial)
library(alluvial)

A_col <- "firebrick3"
B_col <- "darkorange"
C_col <- "aquamarine2"
D_col <- "dodgerblue2"
E_col <- "darkviolet"
F_col <- "chartreuse2"
G_col <- "goldenrod1"
H_col <- "gray73"
set.seed(39)

ggplot(df,
       aes(y = Time, axis1 = Activity, axis2 = Category, axis3 = Positions)) +
  geom_alluvium(aes(fill = Positions, color = Positions), 
        width = 4/12, alpha = 0.5, knot.pos = 0.3) +
  geom_stratum(width = 4/12, color = "grey36") +
  geom_text(stat = "stratum", label.strata = TRUE) +
  scale_x_continuous(breaks = 1:3, 
       labels = c("Activity", "Category", "Positions/Movements"), expand = c(.01, .05)) +
  ylab("Time 24 hours") +
  scale_fill_manual(values  = c(A_col, B_col, C_col, D_col, E_col, F_col, G_col, H_col)) +
  scale_color_manual(values = c(A_col, B_col, C_col, D_col, E_col, F_col, G_col, H_col)) +
  ggtitle("Physical Activity during the week and weekend") +
  theme_minimal() +
  theme(legend.position = "none", panel.grid.major = element_blank(), 
        panel.grid.minor = element_blank(), axis.text.y = element_blank(), 
        axis.text.x = element_text(size = 12, face = "bold"))

# I also have this code that I run without pre-choosing the colours.
# I like this one because the flow diagram doesn't have any border.

ggplot(df,
       aes(y = Time, axis1 = Activity, axis2 = Category, axis3 = Positions)) +
  scale_x_discrete(limits = c("Activity", "Category", "Positions/Moviments"), 
       expand = c(.01, .05)) +
  ylab("Time 24 hours") +
  geom_alluvium(aes(fill = Positions), width = 4/12, alpha = 0.5, knot.pos = 0.3) +
  geom_stratum() + geom_text(stat = "stratum", label.strata = TRUE) +
  theme_minimal() +
  ggtitle("Physical Activity during the week and weekend") +
  theme(legend.position = "none", panel.grid.major = element_blank(), 
        panel.grid.minor = element_blank(), axis.text.y = element_blank(), 
        axis.text.x = element_text(size = 12, face = "bold"))

这是可视化：

有三件事我真的做不到：

排序Category周和周末之后，如Working、Non Working、Sleep Week、Leisure 和 Sleep Weekend.
对Positions/Movements进行排序，例如Sitting、Lying、Standing、Moving、Stairs、Walk Slow、Walk Fast 和 Running。另外，我想用与流程图相同的颜色填充此列的方块。还有就是有些名字没有足够的space，我不知道是否可以重新设置space来容纳它们，或者把它们放在外面用箭头指示所属的方块给他们。差点忘了，有没有办法手动给每个变量分配颜色，比如color black for Walk Slow？另外，如果可能的话，我想去掉流程图边缘的线条。
有没有办法叠加名称位置和运动？

有什么方法可以改进此可视化并使其美观？

提前致谢，路易斯

Answer 1

这是解决您的一些问题的解决方案。

df <- read_csv('Desktop/plot_alluvial_category_position_plus_moviments.csv')
positions <- c("Sitting", "Lying", "Standing", "Moving", "Stairs", "Walk Slow",
               "Walk Fast", "Running")
df$Positions <- factor(df$Positions, levels = positions, labels = positions)
category <- c("Working", "Non Working", "Sleep Week", "Leisure", 
              "Sleep Weekend")
df$Category <- factor(df$Category, levels = category, labels = category)

ggplot(df,
       aes(y = Time, axis1 = Activity, axis2 = Category, axis3 = Positions)) +
  geom_alluvium(aes(fill = Positions), 
                width = 4/12, alpha = 0.5, knot.pos = 0.3) +
  geom_stratum(width = 4/12, color = "grey36") +
  geom_text(stat = "stratum", label.strata = TRUE, min.height=100) +
  scale_x_continuous(breaks = 1:3, 
                     labels = c("Activity", "Category", "Positions\nMovements"), expand = c(.01, .05)) +
  ylab("Time 24 hours") +
  scale_fill_manual(values  = c(A_col, B_col, C_col, D_col, E_col, F_col, G_col, H_col)) +
  scale_color_manual(values = c(A_col, B_col, C_col, D_col, E_col, F_col, G_col, H_col)) +
  ggtitle("Physical activity during the week and weekend") +
  theme_minimal() +
  theme(legend.position = "none", panel.grid.major = element_blank(), 
        panel.grid.minor = element_blank(), axis.text.y = element_blank(), 
        axis.text.x = element_text(size = 12, face = "bold"))

要对层进行排序，您需要将 Category 和 Position 列转换为设置级别顺序的因子。
要去除流程图的边缘，从您的 aes 级别去除 color = Position 就足够了。
您可以通过在标签中添加换行符来堆叠名称 Position 和 Movement。
您可以将颜色分配给层，但前提是类别始终相同（查看 ggalluvial 文档中的一些示例）。
为了避免小层重叠，可以使用ggalluvial版本0.9.2中引入的geom_text中的min.height参数，如图here.

使用R对Sankey/Alluvial图中的部分变量进行美化排序

Beautifying and sorting some variables in the Sankey/Alluvial diagram using R

charts

r

ggplot2

graph-visualization

sankey-diagram