R合并具有等于最小值的行并保留一列的第一个元素和另一列的最后一个元素

Question

我有一个数据框，其中每一行都有一个 ID 和活动值。下面是一个示例数据框：

test=data.frame(
start=c(1,50,100,150,200,250,300,350,400,450,500,100,150,200),
end=c(50,100,150,200,250,300,350,400,450,500,550,150,200,250),
sub_id=c("subid_1_1","subid_1_2","subid_1_3","subid_1_4","subid_1_5","subid_1_6","subid_1_7","subid_1_8","subid_1_9","subid_1_10","subid_1_11","subid_2_1","subid_2_2","subid_2_3"),
id=c(rep("id_1",11),rep("id_2",3)),
activity=c(-0.2,-0.6,-1,-1.2,-1.6,-1.6,-0.5,0.2,-1.2,-0.8,0.1,0.1,1.2,0.5))




print(test)
       start end     sub_id   id activity
    1      1  50  subid_1_1 id_1     -0.2
    2     50 100  subid_1_2 id_1     -0.6
    3    100 150  subid_1_3 id_1     -1.0
    4    150 200  subid_1_4 id_1     -1.2
    5    200 250  subid_1_5 id_1     -1.6
    6    250 300  subid_1_6 id_1     -1.6
    7    300 350  subid_1_7 id_1     -0.5
    8    350 400  subid_1_8 id_1      0.2
    9    400 450  subid_1_9 id_1     -1.2
    10   450 500 subid_1_10 id_1     -0.8
    11   500 550 subid_1_11 id_1      0.1
    12   100 150  subid_2_1 id_2      0.1
    13   150 200  subid_2_2 id_2      1.2
    14   200 250  subid_2_3 id_2      0.5

对于具有相同 ID 的每一行：

我合并了所有未由值 > -1 分隔且接近最小活动值的行。并保留合并行的 start 的第一个值和 end 的最后一个值，并在所选行上计算平均值。

我是这样做的：

library(dplyr)

threshold <- -1

test.group <- test %>%
  mutate(grp = cumsum(activity > threshold)) 

print(test.group, row.names = F)

test.result <- 
  test.group %>%
  subset(activity <= -1) %>%
  group_by(id, grp) %>%
  arrange(activity) %>%
  summarise(
    start.min = first(start),
    end.min = first(end),
    sub_id.min = first(sub_id),
    activity.min = first(activity),
    start = min(start), 
    end = max(end), 
    activity = mean(activity)
  ) %>% 
  ungroup() %>%
  select(start, end, id, activity, start.min, end.min, activity.min)

我的问题是我不能有连续的 min(activity) 的 min(start.min) 和 max(end.min)。 我总是显示一个地区的最小活动。但是如果我有 2 等于最小值，我该怎么做呢？

预期输出：

start   end id    activity start.min end.min activity.min
  <dbl> <dbl> <fct>    <dbl>     <dbl>   <dbl>            <dbl>
1   100   300 id_1      -1.2       200     300          -1.6
2   400   450 id_1      -1.2       400     450          -1.2

Answer 1

我们可以使用data.table::rleid 创建组，删除activity <= -1 所在的行并在每个组中查找相关数字。

library(dplyr)

test %>%
  group_by(gr = data.table::rleid(activity <= -1)) %>%
  filter(activity <= -1) %>%
  summarise(start_1 = first(start), 
            end_1 = last(end), 
            id = first(id),
            activity_1 = mean(activity), 
            activity.min = min(activity), 
            start.min = min(start[activity == activity.min]), 
            end.min = max(end[activity == activity.min])) %>%
   select(-gr)

#   start_1 end_1 id    activity_1 activity.min start.min end.min
#    <dbl> <dbl> <fct>      <dbl>        <dbl>     <dbl>   <dbl>
#1     100   300 id_1       -1.35         -1.6       200     300
#2     400   450 id_1       -1.2          -1.2       400     450

Answer 2

library(dplyr)

test %>%

  mutate(separated = activity <= -1,
         group = cumsum(c(1, diff(separated) != 0))
         ) %>%

  filter(separated) %>%

  group_by(id,group) %>%
  mutate(avgact = mean(activity),
         minact = min(activity),
         start0 = ifelse(activity == minact,start,NA),
         end0 = ifelse(activity == minact,end,NA)
         ) %>%

  summarise(start = first(start),
            end = last(end),
            act = mean(activity),
            start.min = min(start0, na.rm=T),
            end.min = max(end0, na.rm=T),
            activity.min = min(activity)
            ) %>%

  rename(activity = act)


# A tibble: 2 x 8
# Groups:   id [1]
#   id    group start   end activity start.min end.min activity.min
#   <fct> <dbl> <dbl> <dbl>    <dbl>     <dbl>   <dbl>        <dbl>
# 1 id_1      2   100   300    -1.35       200     300         -1.6
# 2 id_1      4   400   450    -1.2        400     450         -1.2