jeshoots-com-219386-unsplash

程式與統計資料視覺化

ggplot2 theme : 3 elements – text, line, rectangle | R 資料視覺化 – part2

發表於 2018-09-262020-07-03 jamleecute

使用ggplot2套件完成基本的資料視覺化後，就是調整整體繪圖的外觀色彩，此時主要使用的圖層就是 ggplot2 theme()，來調整非資料面的繪圖的三大元素(3 elements)：文字(text elements)、線條(line elements)、方形(rectangle elements)。

ggplot2 theme 3 elements

text elements: plot titles, axis title, legend title, text, axis tick mark labels, 等。
line elements: axis lines, minor and major grid lines, plot panel border, 等。
rectangle elements: plot background, panel background, legend background, 等。

而在theme()圖層中我們用來調整3元素的主要函數如下：

element_text()
element_line()
element_rect()

以下我們將使用R ggplot套件中的txhousing資料集進行示範。txhousing為 TAMU real estate center (http://recenter.tamu.edu/)房仲中心收集有關Texus德州房市的資料。資料共8602個觀測值和9個變數，包括city, year, month, date, sales(number of sales, 銷售量), volume(total value of sales, 總銷售金額), median(Median sale price, 銷售額中位數), listings(Total active listings, 總有效房源), inventory(“month inventory”,表示銷售房源所需的月份時間)。

Step 1: 繪製資料圖層

載入所需套件。

library(ggplot2)
library(ggpubr) #ggarrange()

1 2	library(ggplot2) library(ggpubr) #ggarrange()

先繪製各時間點各城市的房市銷售量圖。

data(txhousing)
df <- txhousing
g1 <-
  ggplot(data = df) +
  geom_line(mapping = aes(x = date, y = sales, group = city, colour = city))
g1

1

2

3

4

5

6

data(txhousing)

df

g1

ggplot(data = df) +

geom_line(mapping = aes(x = date, y = sales, group = city, colour = city))

g1

不顯示圖例。使用theme圖層的legend.position，值可以是(“none”, “left”, “right”, “bottom”, “top”, or two-element numeric vector)。若要不顯示，即指定值為”none”即可。

g2 <- 
  ggplot(data = df) +
  geom_line(mapping = aes(x = date, y = sales, group = city, colour = city)) +
  theme(legend.position = "none")
g2

1

2

3

4

5

g2

ggplot(data = df) +

geom_line(mapping = aes(x = date, y = sales, group = city, colour = city)) +

theme(legend.position = "none")

g2

Step 2: 套用快速佈景主題

我們試使用幾個ggplot2內建的快速佈景主題函數。*因為主題內建會顯示圖例，為了使不顯示圖例，我們亦另外使用theme(legend.position = “none”)。

theme_gray(): panel.background底色為深灰。
theme_bw(): panel.border有上色，panel.background則為白色。

p1 <- g2 + theme_gray() + theme(legend.position = "none") + ggtitle(label = "theme_gray")
p2 <- g2 + theme_bw() + theme(legend.position = "none") + ggtitle(label = "theme_bw")
ggarrange(p1, p2, nrow = 1, ncol = 2)

1

2

3

p1

p2

ggarrange(p1, p2, nrow = 1, ncol = 2)

theme_linedraw(): panel.border有邊框。
theme_light(): panel.border邊框為淡色。

p1 <- g2 + theme_linedraw() + theme(legend.position = "none")+ ggtitle(label = "theme_linedraw")
p2 <- g2 + theme_light() + theme(legend.position = "none")  + ggtitle(label = "theme_light")
ggarrange(p1, p2, nrow = 1, ncol = 2)

1

2

3

p1

p2

ggarrange(p1, p2, nrow = 1, ncol = 2)

theme_minimal(): 沒有panel.border。
theme_classic(): 只剩axis.line。沒有panel.grid.major, panel.grid.minor格線。

p1 <- g2 + theme_minimal() + theme(legend.position = "none") + ggtitle(label = "theme_minimal")
p2 <- g2 + theme_classic() + theme(legend.position = "none") + ggtitle(label = "theme_classic")
ggarrange(p1, p2, nrow = 1, ncol = 2)

1

2

3

p1

p2

ggarrange(p1, p2, nrow = 1, ncol = 2)

theme_void(): 拿去所有panel格線(grid)與邊框(border)
theme_dark(): 深色panel.background。

p1 <- g2 + theme_void() + theme(legend.position = "none") + ggtitle(label = "theme_void")
p2 <- g2 + theme_dark() + theme(legend.position = "none") + ggtitle(label = "theme_dark")
ggarrange(p1, p2, nrow = 1, ncol = 2)

1

2

3

p1

p2

ggarrange(p1, p2, nrow = 1, ncol = 2)

這幾個快速佈景主題函數theme_xx()亦可搭配

base_size : 基本的字體大小（可改變所有圖形上的文字元素）
base_family : 基本字體family

p1 <- g2 + theme_bw() + theme(legend.position = "none")+ ggtitle(label = "base_size unchanged")
p2 <- g2 + theme_bw(base_size = 18) + theme(legend.position = "none")+ ggtitle(label = "base_size = 18")
ggarrange(p1, p2, nrow = 1, ncol = 2)

1

2

3

p1

p2

ggarrange(p1, p2, nrow = 1, ncol = 2)

Step 3: 更改plot panel的背景色彩(background colours)與格線(grid lines)

panel的背景色彩屬性為panel.background。我們可以使用以下element_rect()中的幾個屬性來調整：

fill: 填充顏色。
colour: 外框線條顏色。
size: 外框線條粗細。
linetype: 外框線條樣式(blank, solid, dashed, dotted, dotdash, longdash, twodash)

g2 + theme(panel.background = element_rect(fill = "snow",colour = "mediumorchid",size = 0.5,linetype = "solid"))

1	g2 + theme(panel.background = element_rect(fill = "snow",colour = "mediumorchid",size = 0.5,linetype = "solid"))

panel的背景主要、次要格線屬性分別為panel.grid.major, panel.grid.minor。我們則使用element_line()中的幾個屬性來調整：

colour: 格線顏色。
size: 格線粗細。
linetype: 格線樣式。

g2 + theme(panel.background = element_rect(fill = "lightsteelblue",colour = "lightsteelblue",size = 0.5,linetype = "solid"),
           panel.grid.major = element_line(colour = "white",size = 0.5,linetype = "solid"),
           panel.grid.minor = element_line(colour = "white",size = 0.25,linetype = "dashed"))

1

2

3

g2 + theme(panel.background = element_rect(fill = "lightsteelblue",colour = "lightsteelblue",size = 0.5,linetype = "solid"),

panel.grid.major = element_line(colour = "white",size = 0.5,linetype = "solid"),

panel.grid.minor = element_line(colour = "white",size = 0.25,linetype = "dashed"))

Step 4: 移除panel的邊界border和grid lines格線

如果要移除元件，則可使用element_blank()函數。

g2 + theme(panel.grid.major = element_blank(),
           panel.grid.minor = element_blank(),
           panel.border = element_blank())

1

2

3

g2 + theme(panel.grid.major = element_blank(),

panel.grid.minor = element_blank(),

panel.border = element_blank())

如要留下兩座標軸，則可以調整axis.line屬性。

g2 + theme(panel.grid.major = element_blank(),
           panel.grid.minor = element_blank(),
           panel.border = element_blank(),
           axis.line = element_line(colour = "lightgray",size = 0.5,linetype = "solid"))

1

2

3

4

g2 + theme(panel.grid.major = element_blank(),

panel.grid.minor = element_blank(),

panel.border = element_blank(),

axis.line = element_line(colour = "lightgray",size = 0.5,linetype = "solid"))

Step 5: 調整繪圖背景plot background（不是panel background)

改變plot.background的屬性。並移除panel border保留兩軸線。

g2 + theme(plot.background = element_rect(fill = "pink"),
           panel.border = element_blank(),
           axis.line = element_line(colour = "grey",size = 0.4,linetype = "solid"))

1

2

3

g2 + theme(plot.background = element_rect(fill = "pink"),

panel.border = element_blank(),

axis.line = element_line(colour = "grey",size = 0.4,linetype = "solid"))

Step 6: 使用客製化佈景主題custom theme

ggthemes套件內有多組客製化佈景主題themes和scales，可以一次性改變你的繪圖外觀樣貌。

更多詳細的ggthemes套件說明請參考：https://jrnold.github.io/ggthemes/reference/index.html

install.packages("ggthemes") 
library(ggthemes)

1 2	install.packages("ggthemes") library(ggthemes)

我們來試試幾組佈景主題theme_xx()搭配對應的scale_yy_xx()(yy: color, fill, shape, linetype) 或是綜合color,shape,linetype的組合等。

6-1. Themes based on Stata graph schemes

使用theme_stata()佈景主題，並調整字體大小。

p <- ggplot(mtcars, aes(x = wt, y = mpg, colour = factor(gear))) +
  geom_point()+ 
  facet_wrap(facets = ~ am)

p + theme_stata(base_size = 16) +
  ggtitle("mtcars data")

1

2

3

4

5

6

p

geom_point()+

facet_wrap(facets = ~ am)

p + theme_stata(base_size = 16) +

ggtitle("mtcars data")

加上主題色調。

p + theme_stata(base_size = 16) +
  scale_colour_stata() +
  ggtitle("mtcars data")

1

2

3

p + theme_stata(base_size = 16) +

scale_colour_stata() +

ggtitle("mtcars data")

6-2. Theme with Google Docs Chart defaults

p <- ggplot(mtcars) +
  geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +
  facet_wrap(~am)
p + theme_gdocs() + scale_color_gdocs() + ggtitle("mtcars data")

1

2

3

4

p

geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +

facet_wrap(~am)

p + theme_gdocs() + scale_color_gdocs() + ggtitle("mtcars data")

6-3. Inverse gray theme

p <- ggplot(mtcars) +
  geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +
  facet_wrap(~am)
p + theme_igray()

1

2

3

4

p

geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +

facet_wrap(~am)

p + theme_igray()

6-4. ggplot color theme based on the Economist

p <- ggplot(mtcars) +
  geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +
  facet_wrap(~am) +
  # Economist 將x軸標示在右邊
  scale_y_continuous(position = "right")

#使用economist佈景主題與色調
p + theme_economist() +
  scale_colour_economist()

1

2

3

4

5

6

7

8

9

p

geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +

facet_wrap(~am) +

# Economist 將x軸標示在右邊

scale_y_continuous(position = "right")

#使用economist佈景主題與色調

p + theme_economist() +

scale_colour_economist()

step 7: 自製客製化佈景主題(★★★★)

方法1: 使用theme_set() 函數設定當下R session的佈景主題。

theme_set(theme_gray(base_size = 20)) # not run

1	theme_set(theme_gray(base_size = 20)) # not run

方法2: 你可以自訂函數。舉例：自訂theme_self()。

p <- ggplot(mtcars) +
  geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +
  facet_wrap(~am)

theme_self <- function(base_size = 14, base_family = "黑體-繁 中黑") 
  {
    theme(panel.background = element_rect(fill = "lavender"),
          panel.grid.minor = element_line(linetype = "dashed"),
          text = element_text(family = base_family, size = base_size),
          legend.background = element_rect(fill = "gray96",colour = "gray88"),
          plot.title = element_text(hjust = 0.5), #標題置中
          strip.background = element_rect(fill = "lightgray", colour = NA)
          # axis.line = element_line(colour = "mediumpurple4")
          )
  }

p + theme_self()

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

p

geom_point(aes(x = wt, y = mpg, colour = factor(gear))) +

facet_wrap(~am)

theme_self

{

theme(panel.background = element_rect(fill = "lavender"),

panel.grid.minor = element_line(linetype = "dashed"),

text = element_text(family = base_family, size = base_size),

legend.background = element_rect(fill = "gray96",colour = "gray88"),

plot.title = element_text(hjust = 0.5), #標題置中

strip.background = element_rect(fill = "lightgray", colour = NA)

# axis.line = element_line(colour = "mediumpurple4")

)

}

p + theme_self()

總結

有了對ggplot2 theme()圖層的三元素(text, line, rectangles)有了基本的了解，即可將自己常用的繪圖屬性設定自製成theme_xx()以後便不用重複更改或重複撰寫程式碼，並且可以根據喜好調整，不受現有的佈景主題所限制。

更多R資料視覺化學習筆記：

參考連結：