如何拆解含子DataFrame的复杂DataFrame并保留ResourceId

我正在尝试拆解一个每行包含子DataFrame的复杂DataFrame。问题在于，各列中的子DataFrame尺寸不同（如1x3、2x3和2x2），且子DataFrame的Conversions.Value列存在数值与字符混合的数据格式，拆解时出现can't recycle input of size 3 to size 2.等报错。

DataFrame结构 #

structure(list
(Conversions = list(structure(list(Field = "Volume", 
    Unit = "m3", Value = 338L), class = "data.frame", row.names = 1L), 
    structure(list(Field = "Volume", Unit = "m3", Value = 450L), class = "data.frame", row.names = 1L)),     

Categories = list(structure(list(CategorySystem = c("Base", 
    NA), Title = c("Mineral materials and glass (excluding concrete)", "213.7 Kevytbetoni, Aerated concrete"), ClassificationType = c(NA, 
    "Talo2000")), class = "data.frame", row.names = 1:2), structure(list(
        CategorySystem = c("Base", NA), Title = c("Mineral materials and glass (excluding concrete)", 
        "213.7 Kevytbetoni, Aerated concrete"), ClassificationType = c(NA, 
        "Talo2000")), class = "data.frame", row.names = 1:2)), 
   
DataItems.DataValueItems = list(structure(list(DataModuleCode = c("A1-A3 Conservative",  "A1-A3 Typical"), Value = c(0.43, 0.36)), class = "data.frame", row.names = 1:2), 
        structure(list(DataModuleCode = c("A1-A3 Conservative", 
        "A1-A3 Typical"), Value = c(0.41, 0.34)), class = "data.frame", row.names = 1:2)), 
   
ResourceId = c(7000000995, 7000000996)), row.names = 1:2, class = "data.frame")

已尝试的方法及报错 #

• 方法1：

unnest_wider(df, col = 1:3, names_repair = "universal") 
# 能运行但同一行仍有多个列表形式的观测，且长度不一致

• 方法2：

unnest_longer(df, col = 1:3, names_repair = "universal") %>%
mutate(across(.fns = as.character)) %>%
  type_convert()
# 报错：Can't combine `..1$Conversions$Value` <integer> and `..6$Conversions$Value` <character>.

• 方法3：

df$Conversions=lapply(df$Conversions, FUN=as.character)
unnest_longer(df, col = 1:3, names_repair = "universal") %>%
  mutate(across(.fns = as.character)) %>%
  type_convert()
# 报错：! In row 1, can't recycle input of size 3 to size 2.

补充：单独对各列使用rbindlist可合并，但会丢失唯一标识ResourceId，导致数据无法关联。需要为各列子DataFrame添加对应ResourceId后再合并。

方法1：使用tidyverse工具链（dplyr + purrr） #

核心思路是为每个子DataFrame绑定对应的ResourceId，再分别展开各列，最后按需合并：

1. 处理Conversions列（统一Value类型避免冲突）：

library(dplyr)
library(purrr)

conversions_df <- df %>%
  select(ResourceId, Conversions) %>%
  mutate(Conversions = map(Conversions, ~ .x %>% mutate(Value = as.character(Value)))) %>%
  unnest(Conversions)

2. 处理Categories列：

categories_df <- df %>%
  select(ResourceId, Categories) %>%
  unnest(Categories)

3. 处理DataItems.DataValueItems列（重命名Value避免冲突）：

data_items_df <- df %>%
  select(ResourceId, DataItems.DataValueItems) %>%
  unnest(DataItems.DataValueItems) %>%
  rename(DataValue = Value)

4. 按需合并（按ResourceId关联所有信息）：

combined_df <- conversions_df %>%
  left_join(categories_df, by = "ResourceId") %>%
  left_join(data_items_df, by = "ResourceId")

方法2：使用data.table的rbindlist+map #

利用map2将ResourceId和每个子DataFrame绑定，再用rbindlist合并：

library(data.table)

# 处理Conversions
conversions_dt <- rbindlist(map2(df$ResourceId, df$Conversions, ~ .y %>% 
                                  mutate(ResourceId = .x, Value = as.character(Value))))

# 处理Categories
categories_dt <- rbindlist(map2(df$ResourceId, df$Categories, ~ .y %>% mutate(ResourceId = .x)))

# 处理DataItems.DataValueItems
data_items_dt <- rbindlist(map2(df$ResourceId, df$DataItems.DataValueItems, ~ .y %>% 
                                  mutate(ResourceId = .x))) %>%
  setnames("Value", "DataValue")

# 合并数据集
final_dt <- conversions_dt[categories_dt, on = "ResourceId"][data_items_dt, on = "ResourceId"]

关键说明 #

• 针对Conversions.Value的类型混合问题，提前转为字符型（后续可按需转换为合适类型），避免合并时的类型冲突。
• 通过绑定ResourceId到每个子DataFrame，确保拆分后的数据集保留唯一标识，解决数据关联问题。
• 分列拆分再合并的方式，规避了不同尺寸子DataFrame直接unnest时的回收错误。

内容的提问来源于stack exchange，提问作者schlindd