拒绝采样循环在 R 中产生“长度为零”错误

Question

我收到一个argument is of length zero 错误，我无法找出原因。

此代码应该从一系列数字start 和end 中采样。一些行具有额外的依赖关系，其中某些 ID 需要位于其他 ID 之后。代码通过将列after1 和after2 中的值替换为相应的采样值来检查这些依赖关系。如果不满足所需的依赖关系，则重新采样这些值。

当代码成功运行时，sampled 值应填充满足所需依赖关系的数字。代码针对i 迭代运行，在末尾添加一列表示采样值对应的运行。

我最近调整了在将数据输入 R 之前清理和准备数据的方式。我认为我已经正确转置了代码，但是我收到了旧版本中不存在的以下错误，并且我'不知道如何解决它。我查看了其他帖子，但没有找到适用的解决方案。

Error in if (is.na(filter(dftemp, ID == dftemp[j, k])[6])) { : 
  argument is of length zero
In addition: Warning message:
In as.integer(dftemp[j, k]) : NAs introduced by coercion

这是我目前正在处理的代码：

df <- read_csv("sampling sample set.csv", na = c("#VALUE!", "#N/A", ""))
dftemp <- df
dftemp %>% mutate_if(is.factor, as.character) -> dftemp #change factors to characters

for (i in 1:200){ #determines how many iterations to run

  row_list<-as.list(1:nrow(dftemp))
  q<-0

  while(length(row_list)!=0 & q<10){
    q<-q+1 
    for(j in row_list){ #this loop replaces the check values
      skip_flag<-FALSE #initialize skip flag used to check the replacement sampling
      for(k in 4:5){ #checking the after columns
        if(is.na(dftemp[j,k])){ 
          print("NA break")
          print(i)
          break
        } else if(is.na(as.integer(dftemp[j,k]))==FALSE) { #if it's already an integer, we already did this, next
          print("integer next")
          next
          print("integer next")
        } else if(dftemp[j,k]==""){ #check for blank values
          print("empty string next")
          dftemp[j,k]<-NA #if blank value found, replace with NA
          print("fixed blank to NA")
          next 
        } else if(is.na(filter(dftemp,ID==dftemp[j,k])[6])) { #if the replacement has not yet been generated, move on, but set flag to jump this to the end
          skip_flag<-TRUE
          print("skip flag set")
        } else {
          dftemp[j,k]<-as.integer(filter(dftemp,ID==dftemp[j,k])[6]) #replacing IDs with the sampled dates of those IDs
          print("successful check value grab")
        } #if-else
      } #k for loop
      if(skip_flag==FALSE){
        row_list<-row_list[row_list!=j]
      } else {
        next 
      }
      #sampling section
      if(skip_flag==FALSE){
        dftemp[j,6] <- mapply(function(x, y) sample(seq(x, y), 1), dftemp[j,"start"], dftemp[j,"end"])
        dftemp[j,7]<-i #identifying the run number

        if(any(as.numeric(dftemp[j,4:5])>as.numeric(dftemp[j,6]),na.rm=TRUE)){
          print(j)
          while(any(as.numeric(dftemp[j,4:5])>as.numeric(dftemp[j,6]),na.rm=TRUE)){
            dftemp[j,6] <- mapply(function(x, y) sample(seq(x, y), 1), dftemp[j,"start"], dftemp[j,"end"])
          } #while 
          dftemp[j,7]=i 
        }#if
      }
    } #j for loop
  } #while loop wrapper around j loop
  if(i==1){
    dftemp2<-dftemp
  }else{
    dftemp2<-rbind(dftemp2,dftemp)
  }#else

  #blank out dftemp to prepare for another run
  dftemp<-dftemp
  dftemp$sampled <- NA 
  dftemp %>% mutate_if(is.factor, as.character) -> dftemp 

}#i for loop

这里是示例数据。

structure(list(ID = c("a123-1", "b123-1", "c123-1", "d123-1", 
"e123-1", "f123-1", "g123-1", "h123-1", "i123-1", "j123-1", "k123-1", 
"l123-1", "m123-1", "n123-1"), start = c(-5100, -4760, -4930, 
-4930, -5380, -5280, -4855, -4855, -4855, -4855, -4855, -4855, 
-4810, -4810), end = c(-4760, -4420, -4420, -4420, -5080, -5080, 
-4750, -4750, -4750, -4750, -4750, -4750, -4710, -4710), after1 = c(NA, 
NA, NA, NA, NA, NA, NA, "g123-1", "g123-1", NA, "j123-1", "j123-1", 
NA, NA), after2 = c(NA, NA, NA, NA, NA, NA, NA, NA, "h123-1", 
NA, NA, "k123-1", NA, NA), sampled = c(NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA)), class = c("spec_tbl_df", "tbl_df", 
"tbl", "data.frame"), row.names = c(NA, -14L), spec = structure(list(
    cols = list(ID = structure(list(), class = c("collector_character", 
    "collector")), start = structure(list(), class = c("collector_double", 
    "collector")), end = structure(list(), class = c("collector_double", 
    "collector")), after1 = structure(list(), class = c("collector_character", 
    "collector")), after2 = structure(list(), class = c("collector_character", 
    "collector")), sampled = structure(list(), class = c("collector_logical", 
    "collector"))), default = structure(list(), class = c("collector_guess", 
    "collector")), skip = 1), class = "col_spec"))

Answer 1

正如错误消息所示，问题至少发生在is.na(filter(dftemp,ID==dftemp[j,k])[6]) 行。问题似乎在于 dplyr 的 filter 想要作为输入。考虑以下调用中返回的内容：

#returns a tibble with one value
str(dftemp[8,4])

#returns an empty tibble
filter(dftemp,ID==dftemp[8,4])

#returns True
is.data.frame(filter(dftemp,ID==dftemp[8,4]))

filter 直接想要该值，而不是包含该值的数据框。在您的子集上添加 as.character 应该可以解决此问题。请注意，这可能发生在您的代码中的其他地方，因此您可能需要在其他地方确保您拥有正确的数据类型。下面是一个例子：

 #replace line in question with the following:
 is.na(filter(dftemp,ID==as.character(dftemp[8,4]) )[6])

#testing
if(is.na(filter(dftemp,ID==as.character(dftemp[8,4]) )[6])){print("working")}

#output
[1] "working"