我有姿势血压数据（在3个时间点的脉搏，收缩压和舒张压），以及一些其他关键变量，并且想要创建整齐的血压数据。
我曾尝试对单个变量类型（例如收缩压）使用collect（），但在全部3个变量类型上都不能使用（59个原始参与者，将3个collect（）函数一起组合时变为1593）。
library(tidyverse)
ID <- c(1138, 430, 245, 1431, 1587)
group <- c(1, 1, 2, 3, 4)
measure_1 <- c(8, 14, 6, 11, 28)
systolic_lying <- c(169, 141, 144, 120, 88)
systolic_standing_1 <- c(163, 156, 129, 127, 102)
systolic_standing_3 <- c(179, 152, 146, 127, 106)
diastolic_lying <- c(80, 93, 80, 71, 66)
diastolic_standing_1 <-c(87, 97, 79, 77, 75)
diastolic_standing_3 <-c(92, 100, 83, 79, 77)
pulse_lying <-c(58, 71, 58, 63, 98)
pulse_lying_1 <- c(62, 93, 61, 70, 73)
pulse_lying_3 <- c(67, 97, 63, 71, 76)
bp <- tibble(ID, group, measure_1, systolic_lying, systolic_standing_1, systolic_standing_3, diastolic_lying, diastolic_standing_1, diastolic_standing_3, pulse_lying, pulse_lying_1, pulse_lying_3)
head(bp)
# A tibble: 5 x 12
ID group measure_1 systolic_lying systolic_standi… systolic_standi…
<dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 1138 1 8 169 163 179
2 430 1 14 141 156 152
3 245 2 6 144 129 146
4 1431 3 11 120 127 127
5 1587 4 28 88 102 106
# … with 6 more variables: diastolic_lying <dbl>,
# diastolic_standing_1 <dbl>, diastolic_standing_3 <dbl>,
# pulse_lying <dbl>, pulse_lying_1 <dbl>, pulse_lying_3 <dbl>
仅在其中一个变量上使用聚集，例如
tidybp <- gather(bloodpressure, "systolic_posture", "systolic", systolic_lying:systolic_standing_3)
dim(tidybp)
[1] 15 11 #This is what I expect.
head(tidybp[c(1:3, 10:11)])
# A tibble: 6 x 5
ID group measure_1 systolic_posture systolic
<dbl> <dbl> <dbl> <chr> <dbl>
1 1587 4 28 systolic_lying 88
2 1138 1 8 systolic_standing_1 163
3 430 1 14 systolic_standing_1 156
4 245 2 6 systolic_standing_1 129
5 1431 3 11 systolic_standing_1 127
6 1587 4 28 systolic_standing_1 102
但是，这仍然使所有diastolic_lying，diastolic_standing_1 ... pulse_standing_1 pulse_standing_3仍然散开。
然后我拿起tidybp并用聚会将例如舒张期变量在一起，它使循环收集的收缩数据的df长度增加了三倍。
tidybp <- gather(bp, "systolic_posture", "systolic", systolic_lying:systolic_standing_3)
> tidybp <- gather(tidybp, "diastolic_posture", "diastolic", diastolic_lying:diastolic_standing_3)
> tidybp <- gather(tidybp, "pulse_posture", "pulse", pulse_lying:pulse_standing_3)
> dim(tidybp)
[1] 135 9 #each subsequent iteration triples the number of observations
我的目标是：
tidybp
# A tibble: 15 x 9
ID group measure_1 systolic_posture systolic diastolic_posture diastolic pulse_posture pulse
<dbl> <dbl> <dbl> <chr> <dbl> <chr> <dbl> <chr> <dbl>
1 1138 1 8 systolic_lying 169
2 430 1 14 systolic_lying 141
3 245 2 6 systolic_lying 144
4 1431 3 11 systolic_lying 120
5 1587 4 28 systolic_lying 88
6 1138 1 8 systolic_standing_1 163
7 430 1 14 systolic_standing_1 156
8 245 2 6 systolic_standing_1 129
9 1431 3 11 systolic_standing_1 127
10 1587 4 28 systolic_standing_1 102
11 1138 1 8 systolic_standing_3 179
12 430 1 14 systolic_standing_3 152
13 245 2 6 systolic_standing_3 146
14 1431 3 11 systolic_standing_3 127
15 1587 4 28 systolic_standing_3 106
带有相应的舒张期和脉搏键值对。
不胜感激。
答案 0 :(得分：0)
这可以通过tidyverse来完成，但是使用良好的旧base::reshape，恕我直言要容易得多：
v <- list(c("diastolic_lying", "diastolic_standing_1", "diastolic_standing_3"),
c("pulse_lying", "pulse_lying_1", "pulse_lying_3"),
c("systolic_lying", "systolic_standing_1", "systolic_standing_3"))
reshape(as.data.frame(bp),
v,
direction = "long",
v.names = c("diastolic", "pulse", "systolic"),
times = c("lying", "standing_1", "standing_3"),
idvar = "ID",
timevar = "type") %>%
as_tibble()
# A tibble: 15 x 7
# ID group measure_1 type diastolic pulse systolic
# <dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl>
# 1 1138 1 8 lying 80 58 169
# 2 430 1 14 lying 93 71 141
# 3 245 2 6 lying 80 58 144
# 4 1431 3 11 lying 71 63 120
# 5 1587 4 28 lying 66 98 88
# 6 1138 1 8 standing_1 87 62 163
# 7 430 1 14 standing_1 97 93 156
# 8 245 2 6 standing_1 79 61 129
# 9 1431 3 11 standing_1 77 70 127
# 10 1587 4 28 standing_1 75 73 102
# 11 1138 1 8 standing_3 92 67 179
# 12 430 1 14 standing_3 100 97 152
# 13 245 2 6 standing_3 83 63 146
# 14 1431 3 11 standing_3 79 71 127
# 15 1587 4 28 standing_3 77 76 106
答案 1 :(得分：0)
使用tidyverse并不难实现，而且它不会依赖参数，因此，当数据更改（其他度量）时，代码将继续工作。
library(tidyverse)
ID <- c(1138, 430, 245, 1431, 1587)
group <- c(1, 1, 2, 3, 4)
measure_1 <- c(8, 14, 6, 11, 28)
systolic_lying <- c(169, 141, 144, 120, 88)
systolic_standing_1 <- c(163, 156, 129, 127, 102)
systolic_standing_3 <- c(179, 152, 146, 127, 106)
diastolic_lying <- c(80, 93, 80, 71, 66)
diastolic_standing_1 <-c(87, 97, 79, 77, 75)
diastolic_standing_3 <-c(92, 100, 83, 79, 77)
pulse_lying <-c(58, 71, 58, 63, 98)
pulse_standing_1 <- c(62, 93, 61, 70, 73)
pulse_standing_3 <- c(67, 97, 63, 71, 76)
bp <- tibble(ID, group, measure_1, systolic_lying, systolic_standing_1, systolic_standing_3,
diastolic_lying, diastolic_standing_1, diastolic_standing_3,
pulse_lying, pulse_standing_1, pulse_standing_3)
bp %>%
gather(key, value, -ID, -group,-measure_1) %>%
mutate(posture = str_extract(key,'(?<=[a-z]{1,20}_).+'),
measure = str_extract(key,'^[a-z]+(?=_)')) %>%
group_by(ID, group,measure_1,posture) %>%
select(-key) %>%
spread(measure, value) %>%
arrange(posture)
我确实将pulse_lying_1和...3更改为站立状态。否则，它们将以预期的不同行结尾。
结果：
# A tibble: 15 x 7
# Groups: ID, group, measure_1, posture [15]
ID group measure_1 posture diastolic pulse systolic
<dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl>
1 245 2 6 lying 80 58 144
2 430 1 14 lying 93 71 141
3 1138 1 8 lying 80 58 169
4 1431 3 11 lying 71 63 120
5 1587 4 28 lying 66 98 88
6 245 2 6 standing_1 79 61 129
7 430 1 14 standing_1 97 93 156
8 1138 1 8 standing_1 87 62 163
9 1431 3 11 standing_1 77 70 127
10 1587 4 28 standing_1 75 73 102
11 245 2 6 standing_3 83 63 146
12 430 1 14 standing_3 100 97 152
13 1138 1 8 standing_3 92 67 179
14 1431 3 11 standing_3 79 71 127
15 1587 4 28 standing_3 77 76 106
答案 2 :(得分：0)
编辑：我的原始解决方案使用left_join产生不良影响，制作了许多数据副本。在下面更正。
我认为这是您正在寻找的输出，在此我们可以看到所有三个姿势以及每个姿势的配对值。
在这种情况下，我使用bind_cols将三个键值对组合为您所描述的格式。
select(bp, ID:measure_1, systolic_lying:systolic_standing_3) %>%
gather("systolic_posture", "systolic", -c(ID:measure_1)) %>%
bind_cols(select(bp, ID:measure_1, diastolic_lying:diastolic_standing_3) %>%
gather("diastolic_posture", "diastolic", -c(ID:measure_1)) %>%
select(-c(ID:measure_1))) %>%
bind_cols(select(bp, ID:measure_1, pulse_lying:pulse_lying_3) %>%
gather("pulse_posture", "pulse", -c(ID:measure_1)) %>%
select(-c(ID:measure_1)))
# A tibble: 15 x 9
ID group measure_1 systolic_posture systolic diastolic_posture diastolic pulse_posture pulse
<dbl> <dbl> <dbl> <chr> <dbl> <chr> <dbl> <chr> <dbl>
1 1138 1 8 systolic_lying 169 diastolic_lying 80 pulse_lying 58
2 430 1 14 systolic_lying 141 diastolic_lying 93 pulse_lying 71
3 245 2 6 systolic_lying 144 diastolic_lying 80 pulse_lying 58
4 1431 3 11 systolic_lying 120 diastolic_lying 71 pulse_lying 63
5 1587 4 28 systolic_lying 88 diastolic_lying 66 pulse_lying 98
6 1138 1 8 systolic_standing_1 163 diastolic_standing_1 87 pulse_lying_1 62
7 430 1 14 systolic_standing_1 156 diastolic_standing_1 97 pulse_lying_1 93
8 245 2 6 systolic_standing_1 129 diastolic_standing_1 79 pulse_lying_1 61
9 1431 3 11 systolic_standing_1 127 diastolic_standing_1 77 pulse_lying_1 70
10 1587 4 28 systolic_standing_1 102 diastolic_standing_1 75 pulse_lying_1 73
11 1138 1 8 systolic_standing_3 179 diastolic_standing_3 92 pulse_lying_3 67
12 430 1 14 systolic_standing_3 152 diastolic_standing_3 100 pulse_lying_3 97
13 245 2 6 systolic_standing_3 146 diastolic_standing_3 83 pulse_lying_3 63
14 1431 3 11 systolic_standing_3 127 diastolic_standing_3 79 pulse_lying_3 71
15 1587 4 28 systolic_standing_3 106 diastolic_standing_3 77 pulse_lying_3 76
答案 3 :(得分：0)
使用 tidyr 1.0.0 ，您可以执行以下操作：
library(tidyr)
pivot_longer(bp, -(1:3),names_to = c(".value","posture"), names_pattern="([^_]+)_(.+)")
#> # A tibble: 25 x 7
#> ID group measure_1 posture systolic diastolic pulse
#> <dbl> <dbl> <dbl> <chr> <dbl> <dbl> <dbl>
#> 1 1138 1 8 lying 169 80 58
#> 2 1138 1 8 standing_1 163 87 NA
#> 3 1138 1 8 standing_3 179 92 NA
#> 4 1138 1 8 lying_1 NA NA 62
#> 5 1138 1 8 lying_3 NA NA 67
#> 6 430 1 14 lying 141 93 71
#> 7 430 1 14 standing_1 156 97 NA
#> 8 430 1 14 standing_3 152 100 NA
#> 9 430 1 14 lying_1 NA NA 93
#> 10 430 1 14 lying_3 NA NA 97
#> # ... with 15 more rows