ggplot2：合并geom_line，geom_point和geom_bar的图例

Question

在ggplot2图中，我将geom_line和geom_point与geom_bar合并，并且在将图例合并到一个框中时遇到问题。

基本情节的代码如下。使用的数据进一步下降。

# Packages
library(ggplot2)
library(scales)

# Basic Plot
ggplot(data = df1, aes(x = Year, y = value, group = variable, 
    colour = variable, shape = variable)) + 
geom_line() + 
geom_point(size = 3) + 
geom_bar(data = df2, aes(x = Year, y = value, fill = variable), 
    stat = "identity", alpha = 0.8) + 
ylab("Current Account Transactions (Billion $)") + 
xlab(NULL) +  
theme_bw(14) + 
scale_x_discrete(breaks = seq(1999, 2013, by = 2)) +
scale_y_continuous(labels = dollar, limits = c(-1, 4), 
    breaks = seq(-1, 4, by = .5)) + 
geom_hline(yintercept = 0) + 
theme(legend.key = element_blank(), 
    legend.background = element_rect(colour = 'black', fill = 'white'), 
    legend.position = "top", legend.title = element_blank()) + 
guides(col = guide_legend(ncol = 1), fill = NULL, colour = NULL)

我的目标是将传说合并在一起。出于某种原因，＆＃34;经常账户余额＆＃34;出现在顶级传奇（我不明白为什么），而出口＆＃34;出口＆＃34;和＆＃34;进口＆＃34;传说混淆了黑色背景和缺少形状。

如果我在fill之外取aes，我可以获得＆＃34; Imports＆＃34;和＆＃34;出口＆＃34;显示正确的形状和颜色，没有黑色背景，但是我失去了fill传奇的＆＃34;经常账户余额。＆＃34;

我之前使用过的一些技巧，即使用scale_colour_manual，scale_shape_manual和scale_fill_manual（可能还有scale_alpha）似乎在这里不起作用。让它工作会很好。但请注意，有了这个技巧，据我所知，人们必须手动指定颜色，形状和填充，我真的不想这样做，因为我对默认颜色/形状/填充非常满意。

我通常会这样做，但它不起作用：

library(RColorBrewer)
cols <- colorRampPalette(brewer.pal(9, "Set1"))(3)
last_plot() + scale_colour_manual(name = "legend", values = cols) + 
    scale_shape_manual(name = "legend", values = c(0,2,1)) + 
    scale_fill_manual(name = "legend", values = "darkred") 

在上面我没有指定标签，因为在我的问题中，我将处理大量数据，手动指定标签是不切实际的。我希望ggplot2使用默认标签。出于同样的原因，我想使用默认的颜色/形状/填充。

其他地方已经报道了类似的问题，例如Construct a manual legend for a complicated plot，但我没有设法解决我的问题。

有什么想法吗？

# Data

df1 <- structure(list(Year = structure(c(1L, 2L, 3L, 4L, 5L, 6L, 7L, 
8L, 9L, 10L, 11L, 12L, 13L, 14L, 15L, 1L, 2L, 3L, 4L, 5L, 6L, 
7L, 8L, 9L, 10L, 11L, 12L, 13L, 14L, 15L), .Label = c("1999", 
"2000", "2001", "2002", "2003", "2004", "2005", "2006", "2007", 
"2008", "2009", "2010", "2011", "2012", "2013"), class = "factor"), 
    variable = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
    2L, 2L, 2L, 2L, 2L, 2L), .Label = c("Exports of goods and services", 
    "Imports of goods and services"), class = "factor"), value = c(1.304557, 
    1.471532, 1.345165, 1.31879, 1.409053, 1.642291, 1.895983, 
    2.222124, 2.569492, 2.751949, 2.285922, 2.630799, 2.987571, 
    3.08526, 3.178744, 1.600087, 1.882288, 1.740493, 1.776877, 
    1.930395, 2.276059, 2.641418, 3.028851, 3.288135, 3.43859, 
    2.666714, 3.074729, 3.446914, 3.546009, 3.578998)), .Names = c("Year", 
"variable", "value"), row.names = c(NA, -30L), class = "data.frame")

df2 <- structure(list(Year = structure(1:15, .Label = c("1999", "2000 ", 
"2001", "2002 ", "2003", "2004 ", "2005", "2006 ", "2007", "2008 ", 
"2009", "2010 ", "2011", "2012 ", "2013"), class = "factor"), 
    variable = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L), .Label = "Balance on current account", class = "factor"), 
    value = c(-0.29553, -0.410756, -0.395328, -0.458087, -0.521342, 
    -0.633768, -0.745434, -0.806726, -0.718643, -0.686641, -0.380792, 
    -0.44393, -0.459344, -0.460749, -0.400254)), .Names = c("Year", 
"variable", "value"), row.names = c(NA, -15L), class = "data.frame")

修改

在发表我的问题并阅读斯科特的答案之后，我尝试了另一种方法。它在某些方面更接近期望的结果，但在其他方面更进一步。我们的想法是将数据框合并到一个数据框中，并将颜色/形状/填充传递给第一个aes调用内的ggplot。这样做的问题是我得到了一个不受欢迎的斜线＆＃39;跨越传说。在没有删除所有颜色的情况下，我无法删除斜杠。我提到的这种方法的另一个问题是我需要手动指定一堆东西，而我希望尽可能保留默认值。

df <- rbind(df1, df2)
ggplot(data = df, aes(x = Year, y = value, group = variable, colour = variable, 
    shape = variable, fill = variable)) + 
  geom_line(data = subset(df, variable %in% c("Exports of goods and services", "Imports of goods and services"))) + 
  geom_point(data = subset(df, variable %in% c("Exports of goods and services", "Imports of goods and services")), size = 3) + 
  geom_bar(data = subset(df, variable %in% c("Balance on current account")), aes(x = Year, y = value, fill = variable), 
    stat = "identity", alpha = 0.8) 
cols <- c(NA, "darkgreen", "darkblue")
last_plot() + scale_colour_manual(name = "legend", values = cols) + 
  scale_shape_manual(name = "legend", values = c(32, 15, 17)) + 
  scale_fill_manual(name = "legend", values = c("orange", NA, NA)) + 
  ylab("Current Account Transactions (Billion $)") + 
  xlab(NULL) +  
  theme_bw(14) + scale_x_discrete(breaks = seq(1999, 2013, by = 2)) + 
  scale_y_continuous(labels = dollar, limits = c(-1, 4), breaks = seq(-1, 4, by = .5)) + 
  geom_hline(yintercept = 0) + 
  theme(legend.key = element_blank(), legend.background = element_rect(colour = 'black', fill = 'white'), legend.position = "top", legend.title = element_blank()) + 
  guides(col = guide_legend(ncol = 1)) 

添加+ guides(fill = guide_legend(override.aes = list(colour = NULL)))会删除斜线，但也会删除深绿色/深蓝色（它会保留橙色填充）。

Answer 1

消除＆＃34;经常账户余额＆＃34;从顶部图例中显示，您可以将group，colour和shape美学从父ggplot()次调用中转移到geom_line()和{{1}适当的。这样可以特定地控制哪些美学适用于两个共享变量名称的数据集。

geom_point()

这个答案有一些缺点。为一对夫妇命名：1）保留两个单独的图例，如果您决定不对其进行包装，可以伪装（例如，不要设置ggplot(data = df1, aes(x = Year, y = value)) + geom_line(aes(group = variable, colour = variable)) + geom_point(aes(shape = variable, colour = variable), size = 3) + geom_bar(data = df2, aes(x = Year, y = value, fill = variable), stat = "identity", position = 'identity', alpha = 0.8, guide = 'none') + ylab("Current Account Transactions (Billion $)") + xlab(NULL) + theme_bw(14) + scale_x_discrete(breaks = seq(1999, 2013, by = 2)) + scale_y_continuous(labels = dollar, limits = c(-1, 4), breaks = seq(-1, 4, by = .5)) + geom_hline(yintercept = 0) + guides(col = guide_legend(ncol = 1)) + theme(legend.key = element_blank(), legend.background = element_rect(colour = 'black', fill = 'white'), legend.position = "top", legend.title = element_blank(), legend.box.just = "left") ）。 2）从顶部图例中删除df2变量意味着它不会消耗第一个默认颜色（如前所述，仅仅是巧合），所以现在＆＃34;平衡......＆＃34;和＆＃34;出口......＆＃34;两者都显示为粉红色，因为填充图例会回收默认色标。