如何通过类似的列重塑数据框

Question

我有一个数据框，其中有五列，如下所示：

 id    p1    p2   time                      group
___   ___   ___  ____                      _______
 1     1.2  1.9  2016-10-09 01:00:00         1
 1     1.8  1.3  2016-10-09 03:00:00         1
 1     1.2  1.9  2016-10-09 03:00:00         2
 1     1.8  1.3  2016-10-09 06:00:00         2
 3     1.2  1.9  2016-10-09 09:00:00         1
 3     1.8  1.3  2016-10-09 12:00:00         1

由此，我需要为每个id和每个组将长整形调整为宽幅，如下所示：

 id    group      p1_start    p2_start    time_start           p1_complete p2_complete    time_complete                      
 ___   ______    __________   ________    ___________          ________  ______    __________   ________    
  1          1        1.2        1.9      2016-10-09 01:00:00   1.2        1.9      2016-10-09 03:00:00   
  1          2        1.2        1.9      2016-10-09 06:00:00   1.2        1.9      2016-10-09 03:00:00        
  3          1        1.2        1.9      2016-10-09 09:00:00   1.2        1.9      2016-10-09 12:00:00        

所以我尝试了

reshape(DT, idvar = c("id","group"), timevar = "group", direction = "wide")

但这导致了未预期的输出。

任何帮助表示赞赏。

Answer 1

试试这个， df是您的原始数据。 library(data.table) setDT(df) df <- df[, c(.SD[1,], .SD[2,]), by = c('id', 'group')] names(df) <- c('id', 'group', 'p1_start', 'p2_start', 'time_start', 'p1_complete', 'p2_complete', 'time_complete')

Answer 2

如果您不坚持使用data.table解决方案：

library(dplyr) # for pipes `%>%`
library(tidyr) # for `spread`
df %>%
  cbind(spread_grp = c("start","complete")) %>% # adds column which alternates "start" and "complete"
  nest(p1,p2,time)                          %>% # nest the columns we want to spread
  spread(spread_grp,data)                   %>% # spreads our nested column
  unnest(.sep="_")                              # unnest, concatenating the original colum names with the spread_grp values

#   id group complete_p1 complete_p2       complete_time start_p1 start_p2          start_time
# 1  1     1         1.8         1.3 2016-10-09 03:00:00      1.2      1.9 2016-10-09 01:00:00
# 2  1     2         1.8         1.3 2016-10-09 06:00:00      1.2      1.9 2016-10-09 03:00:00
# 3  3     1         1.8         1.3 2016-10-09 12:00:00      1.2      1.9 2016-10-09 09:00:00

名称与您的预期输出中的名称不完全相同，希望这不是问题。

数据

df <- read.table(text="id    p1    p2   time                      group
1     1.2  1.9  '2016-10-09 01:00:00'         1
                 1     1.8  1.3  '2016-10-09 03:00:00'         1
                 1     1.2  1.9  '2016-10-09 03:00:00'         2
                 1     1.8  1.3  '2016-10-09 06:00:00'         2
                 3     1.2  1.9  '2016-10-09 09:00:00'         1
                 3     1.8  1.3  '2016-10-09 12:00:00'         1",stringsAsFactor = FALSE,header=TRUE)