整理“基准”专栏

Question

我有一个大多数整洁的数据框，但是2列包含基准，而不是将基准作为观察结果。 我如何整理这个，以便“Facility_score”和“TTP”col_names作为每个独特的FYQ和Metric组合的“Facility_label”下的观察结果添加？

输入数据：

library(zoo)

dd <- structure(list(Facility_label = structure(c(1L, 1L, 1L, 1L, 2L, 
2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 
3L, 3L, 3L, 4L, 4L, 4L, 4L, 4L, 4L), .Label = c("A", "B", "C", 
"D", "Nashville"), class = "factor"), FYQ = structure(c(2017.75, 
2018, 2018.25, 2018.5, 2017.75, 2018, 2018.25, 2018.5, 2018.75, 
2017.75, 2018, 2018.25, 2018.5, 2018.75, 2017.75, 2018, 2018.25, 
2018.5, 2018.75, 2017.75, 2018, 2018.25, 2018.5, 2018.75, 2017.75, 
2018, 2018.25, 2018.5, 2018.75, 2017.75), class = "yearqtr"), 
    Metric = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    2L, 2L, 2L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 
    1L, 1L, 1L, 1L, 1L, 2L), .Label = c("Safety Recall", "Turnaround days", 
    "Consult Active <= 30d", "Consult Pending <- 7d", "Consult Scheduling <- 90d", 
    "ICB Compliance Rate", "FCA Assessment", "Minor construction execution", 
    "NRM funding execution", "Deficincies", "%Deficienceis corrected among corrected or action plan", 
    "%Deficienceis corrected or action plan", "Ratio of Hospital Staff to HR FTE", 
    "Turnover Rate", "GEMS no Action", "Lost time claims", "RTMS risk score", 
    "DOC Control", "Loaner deficiencies", "Pretreatment", "RME rate", 
    "SPS staff vacany rate", "Stock Inactive", "MSPV-NG", "Days to close prosthetis consult", 
    "%Prosthetic PO using national contracts"), class = "factor"), 
    Facility_score = c(84.78802993, 95.59659091, 100, 100, 77.61732852, 
    57.87671233, 81.28898129, 33.33333333, 31.57894737, 10.2, 
    7.902356902, 8.62, 11.71, 13.15, 30.98236776, 33.26086957, 
    31.19584055, 54.54545455, 27.27272727, 11, 17.19132653, 26.02008197, 
    22.29, 30.41, 89.09090909, 93.47826087, 82.10735586, 91.66666667, 
    87.5, 3.2), `Facility mean` = c(85.35550152, 87.31899147, 
    93.11498231, 100, 85.35550152, 87.31899147, 93.11498231, 
    100, 100, 12, 13.06073298, 12.2, 11.51, 10.56, 85.35550152, 
    87.31899147, 93.11498231, 100, 100, 12, 13.06073298, 12.2, 
    11.51, 10.56, 85.35550152, 87.31899147, 93.11498231, 100, 
    100, 12), TTP_score = c(100, 100, 100, 100, 100, 100, 100, 
    100, 100, 5.65, 5.063953488, 4.779310345, 4.47, 4.545, 100, 
    100, 100, 100, 100, 5.65, 5.063953488, 4.779310345, 4.47, 
    4.545, 100, 100, 100, 100, 100, 5.65)), row.names = c(NA, 
-30L), class = c("tbl_df", "tbl", "data.frame"))

期望的输出：

dd_output <- structure(list(Facility_label = c("A", "Facility mean", "TTP score", 
"A", "Facility mean", "TTP score", "A", "Facility mean", "TTP score", 
"A", "Facility mean", "TTP score", "B", "B", "B", "B", "B", "B", 
"Facility mean", "TTP score", "B", "Facility mean", "TTP score", 
"B", "Facility mean", "TTP score", "B", "Facility mean", "TTP score", 
"B", "Facility mean", "TTP score", "C", "C", "C", "C", "C", "C", 
"C", "C", "C", "C", "D", "D", "D", "D", "D", "D"), FYQ = c("2017 Q4", 
"2017 Q4", "2017 Q4", "2018 Q1", "2018 Q1", "2018 Q1", "2018 Q2", 
"2018 Q2", "2018 Q2", "2018 Q3", "2018 Q3", "2018 Q3", "2017 Q4", 
"2018 Q1", "2018 Q2", "2018 Q3", "2018 Q4", "2017 Q4", "2017 Q4", 
"2017 Q4", "2018 Q1", "2018 Q1", "2018 Q1", "2018 Q2", "2018 Q2", 
"2018 Q2", "2018 Q3", "2018 Q3", "2018 Q3", "2018 Q4", "2018 Q4", 
"2018 Q4", "2017 Q4", "2018 Q1", "2018 Q2", "2018 Q3", "2018 Q4", 
"2017 Q4", "2018 Q1", "2018 Q2", "2018 Q3", "2018 Q4", "2017 Q4", 
"2018 Q1", "2018 Q2", "2018 Q3", "2018 Q4", "2017 Q4"), Metric = c("Safety Recall", 
"Safety Recall", "safety Recall", "Safety Recall", "Safety Recall", 
"Safety Recall", "Safety Recall", "Safety Recall", "Safety Recall", 
"Safety Recall", "Safety Recall", "Safety Recall", "Safety Recall", 
"Safety Recall", "Safety Recall", "Safety Recall", "Safety Recall", 
"Turnaround days", "Turnaround days", "Turnaround days", "Turnaround days", 
"Turnaround days", "Turnaround days", "Turnaround days", "Turnaround days", 
"Turnaround days", "Turnaround days", "Turnaround days", "Turnaround days", 
"Turnaround days", "Turnaround days", "Turnaround days", "Safety Recall", 
"Safety Recall", "Safety Recall", "Safety Recall", "Safety Recall", 
"Turnaround days", "Turnaround days", "Turnaround days", "Turnaround days", 
"Turnaround days", "Safety Recall", "Safety Recall", "Safety Recall", 
"Safety Recall", "Safety Recall", "Turnaround days"), Facility_score = c(84.78802993, 
85.35550152, 100, 95.59659091, 87.31899147, 100, 100, 93.11498231, 
100, 100, 100, 100, 77.61732852, 57.87671233, 81.28898129, 33.33333333, 
31.57894737, 10.2, 12, 5.65, 7.902356902, 13.06073298, 5.063953488, 
8.62, 12.2, 4.779310345, 11.71, 11.51, 4.47, 13.15, 10.56, 4.545, 
30.98236776, 33.26086957, 31.19584055, 54.54545455, 27.27272727, 
11, 17.19132653, 26.02008197, 22.29, 30.41, 89.09090909, 93.47826087, 
82.10735586, 91.66666667, 87.5, 3.2)), class = c("spec_tbl_df", 
"tbl_df", "tbl", "data.frame"), row.names = c(NA, -48L), spec = structure(list(
    cols = list(X1 = structure(list(), class = c("collector_skip", 
    "collector")), Facility_label = structure(list(), class = c("collector_character", 
    "collector")), FYQ = structure(list(), class = c("collector_character", 
    "collector")), Metric = structure(list(), class = c("collector_character", 
    "collector")), Facility_score = structure(list(), class = c("collector_double", 
    "collector"))), default = structure(list(), class = c("collector_guess", 
    "collector")), skip = 1), class = "col_spec"))

Answer 1

我们可以用tidyr::gather和一些dplyr::mutate操作来做到这tidyr::gather ：

library(tidyverse)

dd %>%
  mutate(ID = row_number()) %>%
  gather(var, Facility_score, Facility_score:TTP_score) %>%
  group_by(FYQ, Metric,
           temp_ID = case_when(var == "Facility mean" ~ 1, 
                               var == "TTP_score" ~ 2, 
                               TRUE ~ 0)) %>%
  slice(if(any(temp_ID == 0)) row_number() else 1) %>%
  mutate(Facility_label = if_else(var == "Facility_score", as.character(Facility_label), var)) %>%
  ungroup() %>%
  arrange(ID, temp_ID) %>% 
  select(ID, everything(), -var, -temp_ID) 

请注意，我已添加ID列以指示原始行号。 这使得当同一行中的所有分数组合时不那么混乱。

笔记：

1. gather从宽到长格式TTP_score ，将Facility_score ， Facility mean和TTP_score的条目视为新的Facility_score 。 创建变量var以临时存储值标签。

2. 然后，我们group_by FYQ ， Metric ，和临时ID变量（ temp_ID通过设置创建） Facility mean到1 ， TTP_score至2 ，一切都在其他var到0 。

3. 基于temp_ID ，我们使用slice来获取所有行（如果它是0 ，否则只使用第一行。 这有效地返回与Facility_score对应的所有行，但在每个FYQ + Metric组合中仅返回Facility mean和TTP_score中的一个。

4. 接下来，我们将Facility_label替换为var的相应标签。

5. 最后， ungroup ， arrange由ID和temp_ID ，并重新排列列的顺序，而删除var和temp_ID （ everything ，当我们想在前面带一个变量（一个或多个），同时保持其他变量不变是很有用的）。

输出：

# A tibble: 50 x 5
      ID Facility_label FYQ           Metric          Facility_score
   <int> <chr>          <S3: yearqtr> <fct>                    <dbl>
 1     1 A              2017 Q4       Safety Recall             84.8
 2     1 Facility mean  2017 Q4       Safety Recall             85.4
 3     1 TTP_score      2017 Q4       Safety Recall            100  
 4     2 A              2018 Q1       Safety Recall             95.6
 5     2 Facility mean  2018 Q1       Safety Recall             87.3
 6     2 TTP_score      2018 Q1       Safety Recall            100  
 7     3 A              2018 Q2       Safety Recall            100  
 8     3 Facility mean  2018 Q2       Safety Recall             93.1
 9     3 TTP_score      2018 Q2       Safety Recall            100  
10     4 A              2018 Q3       Safety Recall            100  
11     4 Facility mean  2018 Q3       Safety Recall            100  
12     4 TTP_score      2018 Q3       Safety Recall            100  
13     5 B              2017 Q4       Safety Recall             77.6
14     6 B              2018 Q1       Safety Recall             57.9
15     7 B              2018 Q2       Safety Recall             81.3
16     8 B              2018 Q3       Safety Recall             33.3
17     9 B              2018 Q4       Safety Recall             31.6
18     9 Facility mean  2018 Q4       Safety Recall            100  
19     9 TTP_score      2018 Q4       Safety Recall            100  
20    10 B              2017 Q4       Turnaround days           10.2
# ... with 30 more rows