根据其他两列的值更改 R 中一列的值

Question

我正在尝试根据另外两列的值更改一列的值。 到目前为止，这让我有点头疼，我不确定这是否可能。

我的数据集看起来像这样。 一列是时间，另外两列反映子代父代关系。 在时间点 1 等奇怪的情况下，我的后代“D”第一次出现在数据集中，并且在上一个时间点还没有同时充当后代和父亲时间。

数据

structure(list(time = c(0L, 0L, 0L, 1L, 1L, 1L, 2L, 2L, 2L), 
    offspring = c("A", "B", "C", "A", "D", "E", "A", "F", "G"
    ), parent = c(NA, NA, NA, "A", "B", "D", "A", "A", "F")), class = "data.frame", row.names = c(NA, 
-9L))

我想要帮助的是

1. 找到存在于一个时间点但不存在于前一个时间点的所有后代（不考虑时间点 0），并像 D 和 F 一样作为后代和父亲

2. 当我找到它们时，我想将一个确切的时间点减少 0.5

time  offspring  parent
 0       A        NA
 0       B        NA
 0       C        NA
 1       A        A
 0.5     D        B 
 1       E        D
 2       A        A
 1.5     F        A
 2       G        F

对此问题的任何帮助或指导将不胜感激。

Answer 1

创建 2 个数据框，查找每只动物作为父母和后代的第一次出现。
找出两个组合列中出现的时间和动物，然后更新原始 dataframe 中的时间。

df <-structure(list(time = c(0L, 0L, 0L, 1L, 1L, 1L, 2L, 2L, 2L), 
                    offspring = c("A", "B", "C", "A", "D", "E", "A", "F", "G"), 
                    parent = c(NA, NA, NA, "A", "B", "D", "A", "A", "F")), class = "data.frame", 
                    row.names = c(NA, -9L))


library(dplyr)
#find the row where each Letter First appears as both a parent and offspring 
parents <-df %>% filter(complete.cases(.)) %>% group_by(parent) %>% slice(1) %>% select(time, parent)
offsprings <- df  %>% group_by(offspring) %>% slice(1) %>% select(time, offspring)

combined <- full_join(offsprings, parents)
#rows where the names match for both parent and offspring
matchingrows <-which(combined$parent == combined$offspring)

#update the times
for (i in matchingrows){
   row = which(df$time == combined$time[i] & df$offspring == combined$offspring[i])
   df$time[row] <- df$time[row] - 0.5
}
df

Answer 2

如果您愿意，请在 data.table 中：

library(data.table)

DT <- data.table(time = c(0,0,0,1,1,1,2,2,2),
                 offspring = c('A', 'B', 'C', 'A', 'D', 'E', 'A', 'F', 'G'),
                 parent = c(NA, NA, NA, 'A', 'B', 'D', 'A', 'A', 'F'))

for (i in seq_len(nrow(DT))) {
  DT[i, time := fifelse(time != 0 & offspring %chin% DT[, parent] & !(offspring %chin% DT[seq_len(i-1), offspring]),
                        time - 0.5,
                        time)]
}

> DT
   time offspring parent
1:  0.0         A   <NA>
2:  0.0         B   <NA>
3:  0.0         C   <NA>
4:  1.0         A      A
5:  0.5         D      B
6:  1.0         E      D
7:  2.0         A      A
8:  1.5         F      A
9:  2.0         G      F

与 dplyr：

library(dplyr)
library(tibble)

tbl <- tibble(time = c(0,0,0,1,1,1,2,2,2),
              offspring = c('A', 'B', 'C', 'A', 'D', 'E', 'A', 'F', 'G'),
              parent = c(NA, NA, NA, 'A', 'B', 'D', 'A', 'A', 'F'))

for (i in seq_len(nrow(tbl))) {
  tbl[i,][['time']] <- tbl[i, ] %>% mutate(time = if_else(time != 0 &
                                         offspring %in% tbl[['parent']] &
                                         !(offspring %in% tbl[seq_len(i-1),][['offspring']]),
                                       time - 0.5,
                                       time)) %>% pull(time)
}

> tbl
# A tibble: 9 x 3
   time offspring parent
  <dbl> <chr>     <chr> 
1   0   A         NA    
2   0   B         NA    
3   0   C         NA    
4   1   A         A     
5   0.5 D         B     
6   1   E         D     
7   2   A         A     
8   1.5 F         A     
9   2   G         F   

Answer 3

我的解决方案可能不是最简洁的，但我能够使它工作并且它可以推广到更大的数据集。 我确信有办法改进这一点，所以我很想看看其他人想出什么。 首先，我遇到了 0 个下标的问题，所以我在最后减去的时间列中添加了 2 个。

这个想法是我遍历行，我找到了当年（第 0 年之后）但不是前一年的后代的个体。 然后我检查了哪些人在当年也是父母。 我将这些人在那个时间段内作为后代的行编译成一个向量，因为我们稍后会删除它们。 然后，我用 time-.5、那个后代和它的父母创建一个新行。 我将它们编译成一个新的数据框，它将替换被删除的行。

因为每个时间戳都有重复，所以我使要删除的行向量和要添加的行的 df 唯一。 然后我对原始数据框进行删除和添加，并让数据类型一致。

    parent_offspring <- data.frame(
                 "time" = c( rep(0,3), rep(1,3), rep(2,3)),
                 "offspring" = c("A","B","C","A","D","E","A","F","G"),
                 "parent" = c(NA, NA, NA, "A","B","D","A","A","F")
          )

    po<- parent_offspring
    po$time <- po$time+2
    delete_vec <- vector()
    df_to_add <- data.frame()


    for (i in seq_along(po$time)) {
    q <- po$time[[i]] # Value of "Time" variable for the row
    a <- which(po$time == q) # Rows sharing that value of "Time"
    offspring_curr <- po$offspring[a]  # Offspring at that time
    b <- which(po$time==(q-1))  # Rows of offspring at Time-1
    offspring_prev <- po$offspring[b] # Identities of offspring at Time-1
    f<- offspring_curr[offspring_curr %in% offspring_prev == FALSE] # Which offspring at Time were not offspring at Time-1

    if (length(f) == 0) {
    next
    } else { ##skip ahead if none of the offspring at Time were not offspring at Time - 1

    parents_curr <- po$parent[which(po$time == q)] # Parents at current time
    parent_and_offpsring_curr <- intersect(f,parents_curr) # Which individuals are both parents and offspring at the current time

    if (length(parent_and_offpsring_curr) == 0) {
    next
    } else { ## skip ahead if no individuals are both parents and offspring

    g<- which(po$time==q & po$offspring==parent_and_offpsring_curr) # which offspring row is occupied by an individual who is both a parent and offspring at the current time
    delete_vec <- append(delete_vec,g) #we'll be deleting those rows in the end so we'll keep track of them and save them in a vector
    h<- po$parent[g] # this is the parent for the offspring/parent individual in the current time. 
    add_row<-c((q-.5), parent_and_offpsring_curr, h) # make a new row with the fractional time, parent/offspring individual, and their parent for row when the parent/offspring individual is an offspring
    df_to_add <- rbind(df_to_add,add_row) ## we'll add these rows at the end
  }
}
 }
 
    delete_vec<-unique(delete_vec) ## iteration gave us duplicates
    df_to_add <- unique(df_to_add) ## same as above
    colnames(df_to_add) <- colnames(po) ## fix column names for new df
    po<- po[-delete_vec,] ## remove the offspring rows for the parent/offspring individuals
    po<-rbind(po,df_to_add) ## add the rows with fractional times
    rownames(po) <- c(1:nrow(po)) ## fix the row numbers
    po$time<- as.numeric(po$time) ## time was converted to character when put into a vector with letters
    po$time <- po$time-2 ## back to the original time values

    po

        time offspring parent
      1  0.0         A   <NA>
      2  0.0         B   <NA>
      3  0.0         C   <NA>
      4  1.0         A      A
      5  1.0         E      D
      6  2.0         A      A
      7  2.0         G      F
      8  0.5         D      B
      9  1.5         F      A

然后，您可以使用 dplyr::arrange 按时间升序排列行