R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

huangapple
go评论134阅读模式
英文:

R ggplot2: Replacing aes_string() with aes() and .data[[varname]] when varname may be NULL

问题

我正在更新一些较旧的 R 代码，以便在 aes_string() 被弃用后使用 ggplot2 的 aes()。我承认我不完全理解整个整洁评估的细节，但根据我理解的部分内容，替代 aes_string(shape=shapevar) 的可能是 aes(shape=.data[[shapevar]])，或者可能是 aes(shape=!!sym(shapevar))。

这些替代方法的问题是，当 shapevar 为 NULL 时，它们不起作用，如下面的示例所示：

  1. library('ggplot2')
  3. set.seed(1)
  4. dat <- data.frame(x=rnorm(25),y=rnorm(25),othervar=factor(sample(1:3,25,replace=T)))
  6. make_scatter_string <- function(plotdat,outfile,xvar='x',yvar='y',colorvar=NULL,shapevar=NULL) {
  7.   plt <- ggplot(plotdat,aes_string(x=xvar,y=yvar,color=colorvar,shape=shapevar)) + geom_point()
  8.   ggsave(outfile,plt)
  9. }
  11. make_scatter_data <- function(plotdat,outfile,xvar='x',yvar='y',colorvar=NULL,shapevar=NULL) {
  12.   plt <- ggplot(plotdat,aes(x=.data[[xvar]],y=.data[[yvar]],color=.data[[colorvar]],shape=.data[[shapevar]])) + geom_point()
  13.   ggsave(outfile,plt)
  14. }
  16. make_scatter_sym <- function(plotdat,outfile,xvar='x',yvar='y',colorvar=NULL,shapevar=NULL) {
  17.   plt <- ggplot(plotdat,aes(x=!!sym(xvar),y=!!sym(yvar),color=!!sym(colorvar),shape=!!sym(shapevar))) + geom_point()
  18.   ggsave(outfile,plt)
  19. }
  21. make_scatter_string(dat,'test1.png',colorvar='othervar') # Works, aes_string can handle shapevar being NULL
  22. make_scatter_data(dat,'test2.png',colorvar='othervar') # Gives error message "Error in `.data[[NULL]]`: ! Must subset the data pronoun with a string, not NULL."
  23. make_scatter_sym(dat,'test3.png',colorvar='othervar') # Gives error message "Error in `sym()`: ! Can't convert NULL to a symbol."

如何正确向 aes() 提供类似 shapevar 的变量，该变量可能包含数据中的变量名称，也可能为 NULL？或者是否应该使用其他值来表示 "不要使用此美学"，而不是 NULL？

英文:

I am working on updating some older R code to make it use ggplot2's aes() now that aes_string() is deprecated. I admit I don't understand all of the details of tidy evaluation, but from what I do understand it seems like the replacement for aes_string(shape=shapevar) would be aes(shape=.data[[shapevar]]), or possibly aes(shape=!!sym(shapevar)).

The problem with these replacements is that they do not work when shapevar is NULL, as shown in the example below:

  1. library(&#39;ggplot2&#39;)
  3. set.seed(1)
  4. dat &lt;- data.frame(x=rnorm(25),y=rnorm(25),othervar=factor(sample(1:3,25,replace=T)))
  6. make_scatter_string &lt;- function(plotdat,outfile,xvar=&#39;x&#39;,yvar=&#39;y&#39;,colorvar=NULL,shapevar=NULL) {
  7.   plt &lt;- ggplot(plotdat,aes_string(x=xvar,y=yvar,color=colorvar,shape=shapevar)) + geom_point()
  8.   ggsave(outfile,plt)
  9. }
  11. make_scatter_data &lt;- function(plotdat,outfile,xvar=&#39;x&#39;,yvar=&#39;y&#39;,colorvar=NULL,shapevar=NULL) {
  12.   plt &lt;- ggplot(plotdat,aes(x=.data[[xvar]],y=.data[[yvar]],color=.data[[colorvar]],shape=.data[[shapevar]])) + geom_point()
  13.   ggsave(outfile,plt)
  14. }
  16. make_scatter_sym &lt;- function(plotdat,outfile,xvar=&#39;x&#39;,yvar=&#39;y&#39;,colorvar=NULL,shapevar=NULL) {
  17.   plt &lt;- ggplot(plotdat,aes(x=!!sym(xvar),y=!!sym(yvar),color=!!sym(colorvar),shape=!!sym(shapevar))) + geom_point()
  18.   ggsave(outfile,plt)
  19. }
  21. make_scatter_string(dat,&#39;test1.png&#39;,colorvar=&#39;othervar&#39;) # Works, aes_string can handle shapevar being NULL
  22. make_scatter_data(dat,&#39;test2.png&#39;,colorvar=&#39;othervar&#39;) # Gives error message &quot;Error in `.data[[NULL]]`: ! Must subset the data pronoun with a string, not NULL.&quot;
  23. make_scatter_sym(dat,&#39;test3.png&#39;,colorvar=&#39;othervar&#39;) # Gives error message &quot;Error in `sym()`: ! Can&#39;t convert NULL to a symbol.&quot;

What is the correct way to give aes() a variable like shapevar that may contain the name of a variable in the data or may be NULL? Or is there some other value that should be used to say "don't use this aesthetic" instead of NULL?

答案1

得分: 1

我认为你可以根据is.null(.)的值直接添加个体aes(.)。

  1. make_scatter_data <- function(plotdat, xvar='x', yvar='y', colorvar=NULL, shapevar=NULL) {
  2.   plt <- ggplot(plotdat, aes(x=.data[[xvar]],y =.data[[yvar]])) + geom_point()
  3.   if (!is.null(colorvar)) plt <- plt + aes(color=.data[[colorvar]])
  4.   if (!is.null(shapevar)) plt <- plt + aes(shape=.data[[shapevar]])
  5.   plt
  6. }
  7. make_scatter_data(dat, colorvar='othervar')

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

类似地，

  1. make_scatter_sym <- function(plotdat, xvar='x', yvar='y', colorvar=NULL, shapevar=NULL) {
  2.   plt <- ggplot(plotdat, aes(x=!!sym(xvar), y=!!sym(yvar))) + geom_point()
  3.   if (!is.null(colorvar)) plt <- plt + aes(color = !!sym(colorvar))
  4.   if (!is.null(shapevar)) plt <- plt + aes(shape = !!sym(shapevar))
  5.   plt
  6. }
  7. make_scatter_sym(dat, shapevar='othervar')

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

英文:

I think you can add individual aes(.) directly based on the value of is.null(.).

  1. make_scatter_data &lt;- function(plotdat, xvar=&#39;x&#39;, yvar=&#39;y&#39;, colorvar=NULL, shapevar=NULL) {
  2.   plt &lt;- ggplot(plotdat, aes(x=.data[[xvar]],y =.data[[yvar]])) + geom_point()
  3.   if (!is.null(colorvar)) plt &lt;- plt + aes(color=.data[[colorvar]])
  4.   if (!is.null(shapevar)) plt &lt;- plt + aes(shape=.data[[shapevar]])
  5.   plt
  6. }
  7. make_scatter_data(dat, colorvar=&#39;othervar&#39;)

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

Similarly,

  1. make_scatter_sym &lt;- function(plotdat, xvar=&#39;x&#39;, yvar=&#39;y&#39;, colorvar=NULL, shapevar=NULL) {
  2.   plt &lt;- ggplot(plotdat, aes(x=!!sym(xvar), y=!!sym(yvar))) + geom_point()
  3.   if (!is.null(colorvar)) plt &lt;- plt + aes(color = !!sym(colorvar))
  4.   if (!is.null(shapevar)) plt &lt;- plt + aes(shape = !!sym(shapevar))
  5.   plt
  6. }
  7. make_scatter_sym(dat, shapevar=&#39;othervar&#39;)

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

答案2

得分: 1

以下是您要翻译的内容：

  1. How about:
  3.     make_scatter_data <- function(plotdat, outfile, xvar = x, yvar = y,
  4.                                   colorvar = NULL, shapevar = NULL) {
  5.       plt <- ggplot(plotdat,
  6.                aes(x={{xvar}}, y={{yvar}}, color={{colorvar}}, shape={{shapevar}})) + 
  7.         geom_point()
  8.       ggsave(outfile, plt)
  9.     }
  10.     
  12.     make_scatter_data(dat, "myfile.png")
  14. [![enter image description here][1]][1]
  16.     make_scatter_data(dat, "myfile.png", colorvar = othervar)
  18. [![enter image description here][2]][2]
  20.     make_scatter_data(dat, "myfile.png", shapevar = othervar)
  22. [![enter image description here][3]][3]
  24.     make_scatter_data(dat, "myfile.png", xvar = y, yvar = x, 
  25.                       colorvar = othervar, shapevar = othervar)
  27. [![enter image description here][4]][4]
  30.   [1]: https://i.stack.imgur.com/wUvdN.png
  31.   [2]: https://i.stack.imgur.com/lrVUt.png
  32.   [3]: https://i.stack.imgur.com/7s7Nm.png
  33.   [4]: https://i.stack.imgur.com/B4POd.png
  35. -------
  37. Or if you want to feed in quoted names:
  39.     make_scatter_data <- function(plotdat, outfile, xvar = x, yvar = y,
  40.                                   colorvar = NULL, shapevar = NULL) {
  41.       xvar = sym(xvar)
  42.       yvar = sym(yvar)
  43.       if(!is.null(colorvar)) colorvar = sym(colorvar)
  44.       if(!is.null(shapevar)) shapevar = sym(shapevar)
  45.       
  46.       plt <- ggplot(plotdat,
  47.                     aes(x={{xvar}}, y={{yvar}}, color={{colorvar}}, shape={{shapevar}})) + 
  48.         geom_point()
  49.       plt
  50.       # ggsave(outfile,plt)
  51.     }
  52.     
  53.     make_scatter_data(dat, "myfile.png", xvar = "y", yvar = "x", 
  54.                       colorvar = "othervar")

请注意，代码部分未被翻译。

英文:

How about:

  1. make_scatter_data &lt;- function(plotdat, outfile, xvar = x, yvar = y,
  2.                               colorvar = NULL, shapevar = NULL) {
  3.   plt &lt;- ggplot(plotdat,
  4.            aes(x={{xvar}}, y={{yvar}}, color={{colorvar}}, shape={{shapevar}})) + 
  5.     geom_point()
  6.   ggsave(outfile,plt)
  7. }
  10. make_scatter_data(dat, &quot;myfile.png&quot;)

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

  1. make_scatter_data(dat, &quot;myfile.png&quot;, colorvar = othervar)

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

  1. make_scatter_data(dat, &quot;myfile.png&quot;, shapevar = othervar)

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

  1. make_scatter_data(dat, &quot;myfile.png&quot;, xvar = y, yvar = x, 
  2.                   colorvar = othervar, shapevar = othervar)

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

Or if you want to feed in quoted names:

  1. make_scatter_data &lt;- function(plotdat, outfile, xvar = x, yvar = y,
  2.                               colorvar = NULL, shapevar = NULL) {
  3.   xvar = sym(xvar)
  4.   yvar = sym(yvar)
  5.   if(!is.null(colorvar)) colorvar = sym(colorvar)
  6.   if(!is.null(shapevar)) shapevar = sym(shapevar)
  7.   
  8.   plt &lt;- ggplot(plotdat,
  9.                 aes(x={{xvar}}, y={{yvar}}, color={{colorvar}}, shape={{shapevar}})) + 
  10.     geom_point()
  11.   plt
  12.   # ggsave(outfile,plt)
  13. }
  15. make_scatter_data(dat, &quot;myfile.png&quot;, xvar = &quot;y&quot;, yvar = &quot;x&quot;, 
  16.                   colorvar = &quot;othervar&quot;)

答案3

得分: 1

根据我的回答中的内容，在如何替代已弃用的 ggplot2 函数 aes_string：接受任意数量的命名字符串来指定美学映射？上，@r2evans的一种灵活和通用的方法可能如下所示。

基本上，我将函数的参数重命名为美学的名称，并使用...允许传递额外的美学，也可以与NULL一起使用：

  1. library(ggplot2)
  3. make_scatter_sym <- function(plotdat, outfile, x = "x", y = "y", ...) {
  4.   args <- lapply(list(x = x, y = y, ...), function(x) if (!is.null(x)) sym(x))
  6.   plt <- ggplot(plotdat, aes(!!!args)) +
  7.     geom_point()
  9.   ggsave(outfile, plt)
  11.   plt
  12. }
  14. make_scatter_sym(dat, "test1.png", color = "othervar")
  15. #> Saving 7 x 5 in image

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

  2. make_scatter_sym(dat, "test1.png", color = NULL, shape = NULL)
  3. #> Saving 7 x 5 in image

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]

英文:

Adapting my answer on How to replace the deprecated ggplot2 function aes_string: accepting an arbitrary number of named strings to specify aesthetic mappings?, a flexible and generalized approach of the one by @r2evans may look like so.

Basically I renamed the functions arguments to the names of the aesthetics and used ... to allow to pass additional aesthetics and which also works with NULL:

  1. library(ggplot2)
  3. make_scatter_sym &lt;- function(plotdat, outfile, x = &quot;x&quot;, y = &quot;y&quot;, ...) {
  4.   args &lt;- lapply(list(x = x, y = y, ...), function(x) if (!is.null(x)) sym(x))
  6.   plt &lt;- ggplot(plotdat, aes(!!!args)) +
  7.     geom_point()
  9.   ggsave(outfile, plt)
  11.   plt
  12. }
  14. make_scatter_sym(dat, &quot;test1.png&quot;, color = &quot;othervar&quot;)
  15. #&gt; Saving 7 x 5 in image

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]<!-- -->

  2. make_scatter_sym(dat, &quot;test1.png&quot;, color = NULL, shape = NULL)
  3. #&gt; Saving 7 x 5 in image

R ggplot2: 用 aes() 替代 aes_string()，并在 varname 可能为 NULL 时使用 .data[[varname]]<!-- -->

huangapple
  • 本文由 发表于 2023年5月26日 10:15:35
  • 转载请务必保留本文链接：https://go.coder-hub.com/76337263.html
  • ggplot2
  • r
  • tidyeval
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