使用另一列中的先前信息更改列中的数据

Question

我有一个包含三列的数据集。该列user有两个操作，包括action1和action2。action2仅当action1列有A数据时才包含信息。我想将P数据action1与action2. 例如，如果action2has ac，而下一行有Pin action1，我想P变成Pac，并且会继续（全部P变成Pac）直到action2变化。请注意，此过程应针对每个user.

df<-read.table(text="
user   action1    action2 
1        A          a
1        B          NA
1        P          NA
1        P          NA
1        A          ac
1        P          NA
2        B          NA
2        P          NA
2        A          aa
2        P          NA
2        AB         aa",header=T)

result: (I highlighted those rows that infected)
user   action1    action2 
1        A          a
1        B          NA
1        Pa         NA <-
1        Pa         NA <-
1        A          ac
1        Pac        NA <-
2        B          NA
2        P          NA
2        A          aa
2        Paa        NA <-
2        AB         NA

谢谢

Answer 1

library('data.table')
library('zoo')
# Using zoo::na.locf(), fill NA with the previous value and group by user. Also `na.locf` will not remove NA.
setDT(df)[, V3 := na.locf(action2, na.rm = FALSE), by = .(user)]     
# combine action1 with V3 column if action1 is equal to 'P' and it is not NA.
df[action1 == 'P' & !(is.na(V3)), action1 := paste0(action1, V3)] 
df[, V3 := NULL] # remove V3 column
df
#    user action1 action2
# 1:    1       A       a
# 2:    1       B      NA
# 3:    1      Pa      NA
# 4:    1      Pa      NA
# 5:    1       A      ac
# 6:    1     Pac      NA
# 7:    2       B      NA
# 8:    2       P      NA
# 9:    2       A      aa
# 10:    2     Paa      NA
# 11:    2      AB      aa

数据：

df<-read.table(text="
user   action1    action2 
               1        A          a
               1        B          NA
               1        P          NA
               1        P          NA
               1        A          ac
               1        P          NA
               2        B          NA
               2        P          NA
               2        A          aa
               2        P          NA
               2        AB         aa",header=T, stringsAsFactors = FALSE)