如何使用 R 中的多个列表列展平数据框（从 BigQuery 加载）

Question

我正在使用 R 的 bigrquery 库将数据从 BQ 数据库加载到 R。bigrquery 工作正常，我收到的输出的一个简短示例如下：

dput(my_df_from_bq)
structure(list(season = c("2019", "2019", "2017", "2018", "2018"
), o_or_d = c("Offense", "Defense", "Offense", "Offense", "Defense"
), chances = list(list(num_ato_chances = 6L, ato_pts_scored = 4L, 
    ato_ppp = 0.66667, num_ato_chances_pctile = 0.272955974842767, 
    ato_ppp_pctile = 0.335849056603774), list(num_ato_chances = 7L, 
    ato_pts_scored = 2L, ato_ppp = 0.28571, num_ato_chances_pctile = 0.534591194968553, 
    ato_ppp_pctile = 0.913207547169811), list(num_ato_chances = 5L, 
    ato_pts_scored = 2L, ato_ppp = 0.4, num_ato_chances_pctile = 0.147118921127912, 
    ato_ppp_pctile = 0.177768696362893), list(num_ato_chances = 1L, 
    ato_pts_scored = 0L, ato_ppp = 0, num_ato_chances_pctile = 0, 
    ato_ppp_pctile = 0), list(num_ato_chances = 6L, ato_pts_scored = 8L, 
    ato_ppp = 1.33333, num_ato_chances_pctile = 0.70093839249286, 
    ato_ppp_pctile = 0.165646674826601)), dribbles = list(list(
    dribbles = 928L, dribbles_pctile = 0.437735849056604), list(
    dribbles = 1040L, dribbles_pctile = 0.113207547169811), list(
    dribbles = 771L, dribbles_pctile = 0.0469963220269718), list(
    dribbles = 735L, dribbles_pctile = 0.00489596083231334), 
    list(dribbles = 1049L, dribbles_pctile = 0.103223174214606))), row.names = c(NA, 
-5L), class = c("tbl_df", "tbl", "data.frame")) 

> my_df_from_bq
# A tibble: 5 x 4
  season o_or_d  chances          dribbles        
  <chr>  <chr>   <list>           <list>          
1 2019   Offense <named list [5]> <named list [2]>
2 2019   Defense <named list [5]> <named list [2]>
3 2017   Offense <named list [5]> <named list [2]>
4 2018   Offense <named list [5]> <named list [2]>
5 2018   Defense <named list [5]> <named list [2]>

我从 BQ 加载的 table 包含许多嵌套结构，因此，数据框的这种结构符合预期，因为 bigrquery docs themselves 表明 Nested值变成包含命名列表的列表列。

不过，我现在想把它弄平。您会注意到列表 my_df_from_bq$chances[[1]]$ 的值包括 num_ato_chances、ato_pts_scored、ato_ppp 等。因此，我想展平这个数据框，这样列名就是：

赛季
o_or_d
chances_num_ato_chances
chances_ato_chances_pg
chances_ato_ppp
...
dribbles_dribbles
dribbles_dribbles_pctile

...其中列表名称与每个列表中的值连接在一起。

Answer 1

您可以使用 unnest_wider，但我认为它不会像 open Github issue.

中引用的那样一次性取消嵌套多个列

library(tidyr)
my_df_from_bq %>%
   unnest_wider(chances, names_sep = "_") %>%
   unnest_wider(dribbles, names_sep = "_")

#  season o_or_d chances_num_ato… chances_ato_pts… chances_ato_ppp chances_num_ato…
#  <chr>  <chr>             <int>            <int>           <dbl>            <dbl>
#1 2019   Offen…                6                4           0.667            0.273
#2 2019   Defen…                7                2           0.286            0.535
#3 2017   Offen…                5                2           0.4              0.147
#4 2018   Offen…                1                0           0                0    
#5 2018   Defen…                6                8           1.33             0.701
# … with 3 more variables: chances_ato_ppp_pctile <dbl>, dribbles_dribbles <int>,
#   dribbles_dribbles_pctile <dbl>

如何使用 R 中的多个列表列展平数据框（从 BigQuery 加载）

How to flatten dataframe (loaded from BigQuery) with multiple list-columns in R

r

list

data-manipulation

dataframe

bigrquery