基于前导角色的条件重塑

如何解决基于前导角色的条件重塑

我有两列数据框x：

case_num <- c("ABC-01","111 XX","ABC-02","Y32 XX","113 XX","ABC-03","112 XX","Z90 XX")
transac_num <- c("DEF-01",NA,"DEF-02","DEF-03",NA)

x <- data.frame(case_num,transac_num)

case_num    transac_num
ABC-01      DEF-01
111 XX       
ABC-02      DEF-02
111 XX      
Y32 XX
113 XX
ABC-03      DEF-03
112 XX
Z90 XX

我想重整数据集的格式，以便所有不以“ ABC”开头的行都与位于其上方的case_num放在同一行的单独列中，以“ ABC”开头的行将保留在原处。所需的输出是：

case_num  transac_num     charge1    charge2    charge3
ABC-01    DEF-01          111 XX
ABC-02    DEF-02          111 XX     Y32 XX     113 XX
ABC-03    DEF-03          112 XX     Z90 XX

我尝试通过首先使用case_number字段创建一个新列来指定transac_num是案件还是指控来完成此任务：

x$type <- ifelse(is.na(x$transac_num),"charge","case")

结果

case_num    transac_num    type
ABC-01      DEF-01         case
111 XX                     charge
ABC-02      DEF-02         case
111 XX                     charge
Y32 XX                     charge
113 XX                     charge
ABC-03      DEF-03         case
112 XX                     charge
Z90 XX                     charge

但是，仅在条件pivot_wider下，我无法让tidyr的case_num从x$type == "charge"创建新列。想法？

解决方法

我们可以使用fill用先前的非NA值填充NA元素，slice在第一行之外，按'transac_num'分组后，根据序列创建一个列名列从“ transac_num”开始，并使用pivot_wider将“ long”格式重塑为“ wide”格式

library(dplyr)
library(tidyr)
library(stringr)
library(data.table)
x %>%
    fill(transac_num) %>% 
    group_by(transac_num) %>%
    slice(-1) %>%
    mutate(new_col = str_c('charge',rowid(transac_num))) %>% 
    pivot_wider(names_from =new_col,values_from = case_num)
# A tibble: 3 x 4
# Groups:   transac_num [3]
#  transac_num charge1 charge2 charge3
#  <chr>       <chr>   <chr>   <chr>  
#1 ABC-01      111 XX  <NA>    <NA>   
#2 ABC-02      111 XX  Y32 XX  113 XX 
#3 ABC-03      112 XX  Z90 XX  <NA>   

---

如果要保留另一列，请在分组依据之后创建新列

x %>%
    fill(transac_num) %>%
    group_by(transac_num) %>% 
    mutate(case_num_orig = first(case_num)) %>%
    slice(-1) %>%
    ungroup %>% 
    select(case_num_new = case_num,case_num = case_num_orig,transac_num) %>%
    mutate(new_col = str_c('charge',rowid(transac_num))) %>% 
    pivot_wider(names_from = new_col,values_from = case_num_new)
# A tibble: 3 x 5
#  case_num transac_num charge1 charge2 charge3
#  <chr>    <chr>       <chr>   <chr>   <chr>  
#1 ABC-01   DEF-01      111 XX  <NA>    <NA>   
#2 ABC-02   DEF-02      111 XX  Y32 XX  113 XX 
#3 ABC-03   DEF-03      112 XX  Z90 XX  <NA>