合并数据框列集

时间:2019-01-26 19:50:39

标签: r dataframe dplyr tidyverse tidyr

可能是一个简单的

我有这种类型的data.frame

df <- data.frame(sp1.name = c("sp1.n1",NA,"sp1.n3",NA), sp1.id = c("sp1.id1","sp1.id2",NA,NA),
                 sp2.name = c(NA,NA,"sp2.n3",NA), sp2.id = c(NA,NA,NA,"sp2.id4"),
                 sp3.name = c("sp3.n1",NA,NA,NA), sp3.id = c("sp3.id1",NA,NA,NA))

它由每个“ sp”索引的成对的列组成:sp<index>.namesp<index>.id。在此示例中,索引为1,2,3。

我正在寻找一种方法(可能通过tidyverse)来为每个sp合并其对应的名称和ID列对,其中合并规则为:

  1. if !is.na(sp<index>.name) & !is.na(sp<index>.id) return sp<index>.name
  2. if !is.na(sp<index>.name) & is.na(sp<index>.id) return sp<index>.name
  3. else if is.na(sp<index>.name) & !is.na(sp<index>.id) return sp<index>.id
  4. else return NA

因此,在此示例中,生成的data.frame为:

df <- data.frame(sp1 = c("sp1.n1","sp1.id2","sp1.n3",NA),
                 sp2 = c(NA,NA,"sp2.n3","sp2.id4"),
                 sp3 = c("sp3.n1",NA,NA,NA))

1 个答案:

答案 0 :(得分:2)

您可以这样做:

library(tidyverse)

df %>%
  mutate(rn = row_number()) %>%
  gather(id, value, -rn) %>%
  mutate(idx = gsub("\\..*", "", id)) %>%
  group_by(idx, rn) %>%
  mutate(
    value = case_when(
      any(grepl("name", id) & !is.na(value)) & any( (grepl("id", id) & !is.na(value)) | (grepl("id", id) & is.na(value)) ) ~ value[grepl("name", id)],
      any(grepl("name", id) & is.na(value)) & any(grepl("id", id) & !is.na(value)) ~ value[grepl("id", id)],
      TRUE ~ NA_character_)) %>% 
  distinct(idx, value, rn) %>%
  spread(idx, value)

给予:

# A tibble: 4 x 4
# Groups:   rn [4]
     rn sp1     sp2     sp3   
  <int> <chr>   <chr>   <chr> 
1     1 sp1.n1  NA      sp3.n1
2     2 sp1.id2 NA      NA    
3     3 sp1.n3  sp2.n3  NA    
4     4 NA      sp2.id4 NA