Question

我正在尝试对变量'Size'运行Shapiro Wilks测试，使用我使用ddply进行子集化的数据集（通过变量'Site'和'Category'），但我不断收到错误消息。

以下是我的数据集（d）的示例。我有937个观察点，有9个类别和13个站点：

Site    Genus    Size    Category
Arn01   ACR        4       ACR
Arn01   ACR        7       ACR
Arn02   ACR        3       ACR

我为Shapiro Wilks创建了一个函数：

shap.w <- function(input){ #shapiro wilk test function 
   if(sum(!is.na(input$Size)) > 3 & sum(!is.na(input$Size)) < 5000){
      p <- shapiro.test(input$Size)$p.value
      return(p)}else{return(NA)} }

然后，我尝试使用ddply：

将该函数应用于我的数据子集

sw_test <- ddply(d, .(Site, Category), .fun = shap.w)

但是当我这样做时，我收到一条错误消息：

Error in shapiro.test(input$Size) : all 'x' values are identical

即使他们显然不是。任何帮助/建议将不胜感激。

的ETA输出

dput(d[1:20,]):

> dput(d[1:20,])
structure(list(Site = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), .Label = c("Arn01n", 
"Arn02n", "Arn03n", "Arn04n", "Arn05n", "Arn06n", "Arn07n", "Arn08n", 
"Arn09n", "Arn10n", "Arn11n", "Arn12n", "Arn13n"), class = "factor"), 
Genus = structure(c(2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 30L, 30L, 30L, 30L), .Label = c("ACA", 
"ACR", "AST", "COS", "CYP", "ECH", "FUN", "FVA", "FVT", "GAR", 
"GON", "HEL", "HYD", "ISO", "LEA", "LEO", "LEP", "LOB", "MER", 
"MNT", "MST", "MYC", "PAV", "PBR", "PLA", "PLAT", "POC", 
"POD", "PRE", "PRM", "PRS", "PSA", "SAR", "STY"), class = "factor"), 
Size = c(4, 2, 4, 4, 3, 5, 5, 4, 4, 4, 4, 3, 6, 3, 4, 5, 
2, 3, 3, 6), Category = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 8L, 8L, 8L, 8L), .Label = c("ACR", 
"FAV", "FUN", "HEL", "ISO", "MNT", "POC", "PRM", "PRS"), class = "factor")), 
.Names = c("Site", 
"Genus", "Size", "Category"), row.names = c(NA, 20L), class = "data.frame")`

table(d$Size)

的ETA输出

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 22 23 24 25 26 27 28 29 30 31 33 35 36 37 38 39 14 271 525 548 521 424 201 206 50 357 23 95 36 7 171 11 14 30 4 145 11 21 5 46 4 1 5 1 95 1 2 31 3 1 2 1 40 41 42 43 44 45 46 48 50 51 53 55 56 57 60 62 63 65 66 70 72 75 76 80 82 83 85 88 90 94 95 100 105 110 120 125 80 1 9 3 4 22 1 4 42 1 1 4 1 3 64 3 5 9 4 13 1 2 1 20 2 2 2 1 5 1 2 17 1 2 6 2 128 130 143 150 155 160 180 200 230 300 890 920 1 1 1 1 1 1 1 2 1 1 1 1

Answer 1

请注意，如果您返回NA，则is.numeric会提供FALSE：请尝试is.numeric(NA)查看此内容。

您可以返回NA_real_而不是

is.numeric(NA)
[1] FALSE
is.numeric(NA_real_)
[1] TRUE

它仍然是NA：

is.na(NA_real_)
[1] TRUE

但是，as.numeric也应该解决这个问题（也许会仔细检查输入时函数返回到ddply的内容）

Answer 2

好的，感谢我在评论中收到的帮助，我能够通过更新函数代码来解决这个问题：

shap.w <- function(input){           #shapiro-wilks test function 
if(length(unique((input$Size[!is.na(input)]))) > 3 
& length(unique((input$Size[!is.na(input)])))< 5000 ){
p <- shapiro.test(input$Size)$p.value
return(p)}else{return(NA)} }

这将删除小于3 /大于5000的组合（尽管此数据集中不会有任何大于5,000的组合）。一旦我更新了这个，下一行就没有任何问题了。谢谢大家的帮助！

Shapiro.test＆amp; plyr：所有'x'值都相同

2 个答案: