R::ggplot2::geom_points:如何用饼图交换点? [英] R::ggplot2::geom_points: how to swap points with pie charts?
问题描述
我想绘制二维饼图，以显示每个点的复合组"组成.
到目前为止，我使用标签排斥来标记最高分，但它仍然不是很好.我环顾四周，没有看到我要找的东西.
ggplot(data=aggtmp2,aes(x=cluster,y=x,color=groups,shape=dataset)) +geom_jitter() + facet_grid(datasubset~.) +geom_text_repel(data=aggtmp2[aggtmp2$xnorm>.925,],aes(label=groups),size=2)>str(aggtmp2)'data.frame':562 个观察.共 7 个变量:$ group_name: chr "1_1_D1NF_lewisnegative" "1_1_D1NF_lewisnegative" "1_1_D1NF_lewisnegative" "1_1_D1NF_lewisnegative" ...$ cluster :因子 w/39 个级别 "10of10","1of1",..: 30 24 11 18 25 18 30 11 25 24 ...$ x : 数字 0.591 0.591 0.591 0.591 0.591 ...$ xnorm : num 0.921 0.921 0.921 0.921 0.921 ...$ 组:因子 w/43 个级别 "1_1","1_2","1_3",..: 1 1 1 1 1 2 2 2 2 2 ...$ dataset : Factor w/2 levels "D1NF","D2NF": 1 1 1 1 1 1 1 1 1 1 ...$ datasubset: Factor w/5 levels "all","lewisnegative",..: 2 2 2 2 2 2 2 2 2 2 ...这个答案很接近:
许多用于调整大小和 x,y 比例的数字常量需要通过肉眼进行更改以适合您的数据集.
I would like to plot pie charts in two dimensions to show the composition of each point in terms of their composite "groups."
So far, I'm using label repel to label to highest scoring points but it is still not great. I've looked around and I haven't seen what I'm looking for.
ggplot(data=aggtmp2,aes(x=cluster,y=x,color=groups,shape=dataset)) +
geom_jitter() + facet_grid(datasubset~.) +
geom_text_repel(data=aggtmp2[aggtmp2$xnorm>.925,],aes(label=groups),size=2)
> str(aggtmp2)
'data.frame': 562 obs. of 7 variables:
$ group_name: chr "1_1_D1NF_lewisnegative" "1_1_D1NF_lewisnegative" "1_1_D1NF_lewisnegative" "1_1_D1NF_lewisnegative" ...
$ cluster : Factor w/ 39 levels "10of10","1of1",..: 30 24 11 18 25 18 30 11 25 24 ...
$ x : num 0.591 0.591 0.591 0.591 0.591 ...
$ xnorm : num 0.921 0.921 0.921 0.921 0.921 ...
$ groups : Factor w/ 43 levels "1_1","1_2","1_3",..: 1 1 1 1 1 2 2 2 2 2 ...
$ dataset : Factor w/ 2 levels "D1NF","D2NF": 1 1 1 1 1 1 1 1 1 1 ...
$ datasubset: Factor w/ 5 levels "all","lewisnegative",..: 2 2 2 2 2 2 2 2 2 2 ...
This answer comes close: ggplot use small pie charts as points with geom_point But I'm trying to get this done without facet_grid(). That way I can more naturally show composition in the xy-coordinate space I've set up.
Update:
If you want pie charts in particular, you're better off with package scatterpie, https://cran.r-project.org/web/packages/scatterpie/vignettes/scatterpie.html. My method below works with non-pie charts as well though.
I was curious to see if it could be done, and I'm not sure how flexible this solution is, but here's what I came up with. It's worth stepping through this code chunk line by line, stopping before each %>% pipe to see what it's generating as it goes along. This first chunk generates some data: 5 random X and Y values. Then component labels and their values are generated and bound to the Xs and Ys. Then for proof-of-concept I created an additional column that shows the sum of the components for each X-Y pair.
require(dplyr)
require(ggplot2)
df <- data_frame(x1 = rnorm(5), y1 = rnorm(5)) %>%
group_by(x1, y1) %>%
do(data_frame(component = LETTERS[1:3], value = runif(3))) %>%
mutate(total = sum(value)) %>%
group_by(x1, y1, total)
df
Source: local data frame [15 x 5] Groups: x1, y1, total [5]
x1 y1 component value total
<dbl> <dbl> <chr> <dbl> <dbl>
1 -1.0933810 0.4162150 A 0.920992065 2.1406433
2 -1.0933810 0.4162150 B 0.914163390 2.1406433
3 -1.0933810 0.4162150 C 0.305487891 2.1406433
4 -0.9579912 1.4080922 A 0.006967777 0.3149009
5 -0.9579912 1.4080922 B 0.128341286 0.3149009
6 -0.9579912 1.4080922 C 0.179591852 0.3149009
7 0.5617438 -0.8770998 A 0.233895761 1.2324975
8 0.5617438 -0.8770998 B 0.942843309 1.2324975
9 0.5617438 -0.8770998 C 0.055758395 1.2324975
10 0.9970852 -0.4142704 A 0.035965092 1.4261429
11 0.9970852 -0.4142704 B 0.454193773 1.4261429
12 0.9970852 -0.4142704 C 0.935984062 1.4261429
13 1.2253968 0.3557304 A 0.692594728 2.1289173
14 1.2253968 0.3557304 B 0.972569822 2.1289173
15 1.2253968 0.3557304 C 0.463752786 2.1289173
This chunk takes the first dataframe and for each unique x1-y1-total combination, generates a plain pie chart in a list-column called subplots. Each value in that column is a list of ggplot elements to generate a figure. Then it constructs another column that turns each of those subplots into a custom annotation in a column called subgrobs. These annotations are what can be dropped into a larger figure.
df.grobs <- df %>%
do(subplots = ggplot(., aes(1, value, fill = component)) +
geom_col(position = "fill", alpha = 0.75, colour = "white") +
coord_polar(theta = "y") +
theme_void()+ guides(fill = F)) %>%
mutate(subgrobs = list(annotation_custom(ggplotGrob(subplots),
x = x1-total/4, y = y1-total/4,
xmax = x1+total/4, ymax = y1+total/4)))
df.grobs
Source: local data frame [5 x 5]
Groups: <by row>
# A tibble: 5 × 5
x1 y1 total subplots subgrobs
<dbl> <dbl> <dbl> <list> <list>
1 -1.0933810 0.4162150 2.1406433 <S3: gg> <S3: LayerInstance>
2 -0.9579912 1.4080922 0.3149009 <S3: gg> <S3: LayerInstance>
3 0.5617438 -0.8770998 1.2324975 <S3: gg> <S3: LayerInstance>
4 0.9970852 -0.4142704 1.4261429 <S3: gg> <S3: LayerInstance>
5 1.2253968 0.3557304 2.1289173 <S3: gg> <S3: LayerInstance>
Here it just takes the 5 unique x1-y1-total combinations and plots them as a regular ggplot, but then also adds in the custom annotations we made, which are sized proportional to total. Then a fake legend is constructed from the original dataframe df using blank geom_cols.
df.grobs %>%
{ggplot(data = ., aes(x1, y1)) +
scale_x_continuous(expand = c(0.25, 0)) +
scale_y_continuous(expand = c(0.25, 0)) +
.$subgrobs +
geom_text(aes(label = round(total, 2))) +
geom_col(data = df,
aes(0,0, fill = component),
colour = "white")}
A lot of the numeric constants for adjusting the sizes and x,y scales would need to be changed by eye to fit your dataset.
这篇关于R::ggplot2::geom_points:如何用饼图交换点?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！
