将不相等数组的列转换为R中的单个值列

Question

作为previous question 之后的下一步，假设有多个长度不同的数组列。例如：

Col_A	Col_B	Col_C
[0.1,0.5,0.7]	[1.54E12, 1.54E12, 1.54E12]	[1, 3, 4, 5}

我怎样才能采用这种格式并将其重新格式化为以下在适当的情况下为 Col_A 和 Col_b 提供 NA：

Col_A	Col_B	Col_C
0.1	1.54E12	1
0.5	1.54E12	3
0.7	1.54E12	4
NA	NA	5

此代码适用于所有数组都相等但如果数组不相等则会抛出错误：

library(dplyr)
library(stringr)
library(tidyr)
df  %>% 
   mutate(across(everything(), str_extract_all, "(?<=\\[)[^]]+")) %>% 
   unnest(c(NDVIs, dates)) %>% 
   separate_rows(c(NDVIs, dates), sep=",\\s+", convert = TRUE)

Answer 1

我们可以使用来自splitstackshape的cSplit

library(splitstackshape)
library(data.table)
cSplit(setDT(df)[, lapply(.SD, gsub, pattern = "[][}]", 
    replacement = "")], names(df), sep=",", fixed = FALSE, "long")
#   Col_A    Col_B Col_C
#1:   0.1 1.54e+12     1
#2:   0.5 1.54e+12     3
#3:   0.7 1.54e+12     4
#4:    NA       NA     5

数据

df <- structure(list(Col_A = "[0.1,0.5,0.7]", Col_B = "[1.54E12, 1.54E12, 1.54E12]", 
    Col_C = "[1, 3, 4, 5}"), class = "data.frame", row.names = c(NA, 
-1L))

Answer 2

这里tidyverse 的经验不足，所以这是我使用data.table 的解决方案。我在步骤和结果之间包含了所有内容，以显示发生了什么......

library( data.table )
#create sample data
DT <- fread("Col_A  Col_B   Col_C
[0.1,0.5,0.7]   [1.54E12, 1.54E12, 1.54E12]     [1, 3, 4, 5]")
#            Col_A                       Col_B        Col_C
# 1: [0.1,0.5,0.7] [1.54E12, 1.54E12, 1.54E12] [1, 3, 4, 5]

#melt to long format
ans <- melt( DT, measure.vars = names(DT), variable.factor = FALSE )
#    variable                       value
# 1:    Col_A               [0.1,0.5,0.7]
# 2:    Col_B [1.54E12, 1.54E12, 1.54E12]
# 3:    Col_C                [1, 3, 4, 5]

#remove [] and split the value column using ', ' as sepatator
ans[, value := gsub( "\\[|\\]", "", value ) ]
ans[, paste0( "v", 1:length( tstrsplit(ans$value, "," ) ) ) := 
      lapply( tstrsplit(value, "," ), as.numeric ) ][]
#    variable                     value       v1       v2       v3 v4
# 1:    Col_A               0.1,0.5,0.7 1.00e-01 5.00e-01 7.00e-01 NA
# 2:    Col_B 1.54E12, 1.54E12, 1.54E12 1.54e+12 1.54e+12 1.54e+12 NA
# 3:    Col_C                1, 3, 4, 5 1.00e+00 3.00e+00 4.00e+00  5

#transpose (without value-columns) to get wide format again
transpose( ans[, -"value"], make.names = "variable" )
#    Col_A    Col_B Col_C
# 1:   0.1 1.54e+12     1
# 2:   0.5 1.54e+12     3
# 3:   0.7 1.54e+12     4
# 4:    NA       NA     5