将一列列表转换为一列向量

Question

我有一个data.table，由于某种原因有几列list。我需要找到一种快速方法将这些列转换为具有integer 值的常规向量列。

请参阅下面的可重现示例和我的尝试。

代表

dt = structure(list(fromId = c("89a901291abffff", "89a901295b7ffff", 
      "89a9012809bffff", "89a901285cfffff"), travel_time_p001 = c(45L, 
      43L, 38L, 27L), travel_time_p050 = list(54L, 52L, 44L, 35L), 
      travel_time_p098 = list(62L, 60L, 55L, 44L)), row.names = c(NA, 
      -4L), class = c("data.table", "data.frame"))


head(dt)
>             fromId travel_time_p001 travel_time_p050 travel_time_p098
> 1: 89a901291abffff               45               54               62
> 2: 89a901295b7ffff               43               52               60
> 3: 89a9012809bffff               38               44               55
> 4: 89a901285cfffff               27               35               44

sapply(dt, class)
>     fromId travel_time_p001 travel_time_p050 travel_time_p098 
>"character"        "integer"           "list"           "list" 

class(dt$travel_time_p098[1])
> [1] "list"

我发现这种方法可以修复单个列，但我不确定这是否可以更快地完成。

dt[, travel_time_p098 := Reduce(rbind, travel_time_p098)]

现在，当我尝试为多个列执行此操作时，挑战变得更加困难。这是我尝试过但没有成功的方法。我希望有人能想出一个快速的解决方案，或许可以使用data.table 循环和set()。

# identify which columns need to be transformed.
colpositions <- sapply(dt, class) == 'list'
colnms <- names(dt)[colpositions]

# failed attempt
dt[, colnms := Reduce(rbind, .SD), .SDcols = colnms]

有什么建议吗？

Answer 1

如果list 元素的长度与行数相同，则可以选择循环遍历所有列，使用if/else 和unlist 进行检查

dt1 <- dt[,  lapply(.SD, function(x) if(is.list(x)) unlist(x) else x)]
str(dt1)
#Classes ‘data.table’ and 'data.frame': 4 obs. of  4 variables:
# $ fromId          : chr  "89a901291abffff" "89a901295b7ffff" "89a9012809bffff" "89a901285cfffff"
# $ travel_time_p001: int  45 43 38 27
# $ travel_time_p050: int  54 52 44 35
# $ travel_time_p098: int  62 60 55 44

---

或者没有任何检查

dt[, lapply(.SD, unlist)] 

---

如果我们需要使用set

for(j1 in seq_along(dt)) {
     cl1 <- class(dt[[j1]])
     if(cl1 == 'list') {
          set(dt, i = NULL, j = j1, value = unlist(dt[[j1]]))
        }
  }
   

str(dt)
#Classes ‘data.table’ and 'data.frame': 4 obs. of  4 variables:
# $ fromId          : chr  "89a901291abffff" "89a901295b7ffff" "89a9012809bffff" "89a901285cfffff"
# $ travel_time_p001: int  45 43 38 27
# $ travel_time_p050: int  54 52 44 35
# $ travel_time_p098: int  62 60 55 44