ggraph/igraph显示不正确的标签和分层结构

huangapple
go评论95阅读模式
英文:

ggraph/igraph show uncorrece label and hierarchical structure

问题

ggraph/igraph，以下绘图结果不正确：

  1. DE的数量为4.5，但子圆圈分别为8.8和9（正确数量应为“DE cat_a 3.0，DE cat_b 1.5”）

  2. UK/US没有子圆圈

如何修复？谢谢！

  1. library(ggraph)
  2. library(igraph)
  3. library(tidyverse)
  4. library(viridis)
  6. base <- data.frame(from =c('US','US','UK','UK','DE','DE'),
  7.                    to = c('cat_a','cat_b','cat_a','cat_b','cat_a','cat_b'),
  8.                    sales=c(1,3,5,4,3,1.5))
  10. vetic <- base %>% gather(key='type', value ='dim',c(1:2)) %>%
  11.   group_by(dim) %>% summarise(sales = sum(sales)) 
  13. mdf <- graph_from_data_frame(base,directed = TRUE,vertices = vetic)
  15. ggraph(mdf, layout = 'circlepack', weight=sales) + 
  16.   geom_node_circle(aes(fill = depth)) +
  17.     geom_node_text(size=10,aes(label= paste0(name,'\n',sales)))+
  18.   theme_void()+scale_fill_distiller(palette = "RdPu")

ggraph/igraph显示不正确的标签和分层结构

  2. <details>
  3. <summary>英文:</summary>
  5. `ggraph/igraph`, the result in below plot isn&#39;t correct :
  7.  1)The amount of DE is 4.5, but the sub circles are 8.8 and 9 (the correct amount should be &quot;DE cat_a   3.0, DE cat_b   1.5&quot;)  
  9. 2) there is no sub circle for UK/US
  11. How to fix it ? Thanks!
  20.     library(ggraph)
  21.     library(igraph)
  22.     library(tidyverse)
  23.     library(viridis)
  24.     
  25.     base &lt;- data.frame(from =c(&#39;US&#39;,&#39;US&#39;,&#39;UK&#39;,&#39;UK&#39;,&#39;DE&#39;,&#39;DE&#39;),
  26.                        to = c(&#39;cat_a&#39;,&#39;cat_b&#39;,&#39;cat_a&#39;,&#39;cat_b&#39;,&#39;cat_a&#39;,&#39;cat_b&#39;),
  27.                        sales=c(1,3,5,4,3,1.5))
  28.     
  29.     vetic &lt;- base %&gt;% gather(key=&#39;type&#39;, value =&#39;dim&#39;,c(1:2)) %&gt;% 
  30.       group_by(dim) %&gt;% summarise(sales = sum(sales)) 
  31.     
  32.     mdf &lt;- graph_from_data_frame(base,directed = TRUE,vertices = vetic)
  33.     
  34.     
  35.     ggraph(mdf, layout = &#39;circlepack&#39;, weight=sales) + 
  36.       geom_node_circle(aes(fill = depth)) +
  37.         geom_node_text(size=10,aes(label= paste0(name,&#39;\n&#39;,sales)))+
  38.       theme_void()+scale_fill_distiller(palette = &quot;RdPu&quot;) 
  40. [![enter image description here][1]][1]
  43.   [1]: https://i.stack.imgur.com/WfV1F.png
  45. </details>
  48. # 答案1
  49. **得分**: 3
  51. 问题出在你创建图表的方式上。似乎你想要展示每个国家内的“cat_a”和“cat_b”的数量，但你创建的图表好像将“cat_a”和“cat_b”看作单个节点。如果你想要在每个国家内有一个“cat_a”和一个“cat_b”节点，那么你需要将它们指定为不同的节点。
  53. 例如，我们可以这样做：
  55. ``` r
  56. library(ggraph)
  57. library(igraph)
  58. library(tidyverse)
  59. library(viridis)
  61. base <- data.frame(from =c('US','US','UK','UK','DE','DE'),
  62.                    to = c('cat_a','cat_b','cat_a','cat_b','cat_a','cat_b'),
  63.                    sales=c(1,3,5,4,3,1.5))
  65. base_fixed <- base %>% mutate(to = paste(from, to))

这将给出以下数据框：

  1. base_fixed
  2. #>   from       to sales
  3. #> 1   US US cat_a   1.0
  4. #> 2   US US cat_b   3.0
  5. #> 3   UK UK cat_a   5.0
  6. #> 4   UK UK cat_b   4.0
  7. #> 5   DE DE cat_a   3.0
  8. #> 6   DE DE cat_b   1.5

同样，顶点应该为每个生成的9个节点（3个国家节点，以及每个国家内的两个“cat”节点）具有标签和值：

  1. vertices <- base_fixed %>%
  2.   group_by(from) %>%
  3.   summarize(sales = sum(sales)) %>%
  4.   rbind(base_fixed %>% select(-1) %>% rename(from = to))
  6. vertices
  7. #> # A tibble: 9 x 2
  8. #>   from     sales
  9. #>   <chr>    <dbl>
  10. #> 1 DE         4.5
  11. #> 2 UK         9  
  12. #> 3 US         4  
  13. #> 4 US cat_a   1  
  14. #> 5 US cat_b   3  
  15. #> 6 UK cat_a   5  
  16. #> 7 UK cat_b   4  
  17. #> 8 DE cat_a   3  
  18. #> 9 DE cat_b   1.5

然后，可以按照以下方式绘制图表：

  1. graph_from_data_frame(base_fixed, vertices = vertices) %>%
  2.   ggraph(layout = 'circlepack', weight = sales) + 
  3.   geom_node_circle(aes(fill = depth)) +
  4.   geom_node_text(size = 5, 
  5.                  aes(label = ifelse(depth == 0, '',
  6.                       paste0(substr(name, 4, 20), '\n', sales)))) +
  7.   geom_node_label(size = 6, aes(label = ifelse(depth == 1, NA,
  8.                    substr(name, 1, 2)))) +
  9.   theme_void() +
  10.   scale_fill_distiller(palette = "RdPu", guide = 'none') +
  11.   coord_equal()

ggraph/igraph显示不正确的标签和分层结构

创建于2023-05-15，使用 reprex v2.0.2

英文:

The problem lies with how you are creating your graph. It seems you want to show the amount of cat_a and cat_b within each country, but you are creating the graph as though cat_a and cat_b are each a single node. If you want a cat_a and a cat_b node inside each country, then you need to specify them as distinct nodes.

For example, we can do:

  1. library(ggraph)
  2. library(igraph)
  3. library(tidyverse)
  4. library(viridis)
  6. base &lt;- data.frame(from =c(&#39;US&#39;,&#39;US&#39;,&#39;UK&#39;,&#39;UK&#39;,&#39;DE&#39;,&#39;DE&#39;),
  7.                    to = c(&#39;cat_a&#39;,&#39;cat_b&#39;,&#39;cat_a&#39;,&#39;cat_b&#39;,&#39;cat_a&#39;,&#39;cat_b&#39;),
  8.                    sales=c(1,3,5,4,3,1.5))
  10. base_fixed &lt;- base %&gt;% mutate(to = paste(from, to))

Which gives the following data frame:

  1. base_fixed
  2. #&gt;   from       to sales
  3. #&gt; 1   US US cat_a   1.0
  4. #&gt; 2   US US cat_b   3.0
  5. #&gt; 3   UK UK cat_a   5.0
  6. #&gt; 4   UK UK cat_b   4.0
  7. #&gt; 5   DE DE cat_a   3.0
  8. #&gt; 6   DE DE cat_b   1.5

Similarly, the vertices should have a label and value for each of the resulting 9 nodes (3 country nodes, plus two "cat" nodes within each):

  1. vertices &lt;- base_fixed %&gt;%
  2.   group_by(from) %&gt;%
  3.   summarize(sales = sum(sales)) %&gt;%
  4.   rbind(base_fixed %&gt;% select(-1) %&gt;% rename(from = to))
  6. vertices
  7. #&gt; # A tibble: 9 x 2
  8. #&gt;   from     sales
  9. #&gt;   &lt;chr&gt;    &lt;dbl&gt;
  10. #&gt; 1 DE         4.5
  11. #&gt; 2 UK         9  
  12. #&gt; 3 US         4  
  13. #&gt; 4 US cat_a   1  
  14. #&gt; 5 US cat_b   3  
  15. #&gt; 6 UK cat_a   5  
  16. #&gt; 7 UK cat_b   4  
  17. #&gt; 8 DE cat_a   3  
  18. #&gt; 9 DE cat_b   1.5

The graph could then be drawn as follows:

  1. graph_from_data_frame(base_fixed, vertices = vertices) %&gt;%
  2.   ggraph(layout = &#39;circlepack&#39;, weight = sales) + 
  3.   geom_node_circle(aes(fill = depth)) +
  4.   geom_node_text(size = 5, 
  5.                  aes(label = ifelse(depth == 0, &#39;&#39;,
  6.                       paste0(substr(name, 4, 20), &#39;\n&#39;, sales)))) +
  7.   geom_node_label(size = 6, aes(label = ifelse(depth == 1, NA,
  8.                    substr(name, 1, 2)))) +
  9.   theme_void() +
  10.   scale_fill_distiller(palette = &quot;RdPu&quot;, guide = &#39;none&#39;) +
  11.   coord_equal()

ggraph/igraph显示不正确的标签和分层结构

<sup>Created on 2023-05-15 with reprex v2.0.2</sup>

huangapple
  • 本文由 发表于 2023年5月15日 10:07:23
  • 转载请务必保留本文链接：https://go.coder-hub.com/76250463.html
  • ggraph
  • igraph
  • r
  • tidyverse
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定