在 R 中检测具有多个观察值的行

Posted 2023-02-14

【中文标题】在 R 中检测具有多个观察值的行

【英文标题】：Detecting rows with multiple observations in R

【发布时间】：2022-01-22 22:49:11

【问题描述】：

我有一个这样的数据集。我想识别在“颜色”列中具有多个值的所有观察结果，并将它们替换为“多色”

ID  color1   color2
23   red      NA
44   blue     purple
51   yellow   NA
59   green    orange

像这样：

ID  color   
23   red      
44   multicolor     
51   yellow     
59   multicolor   

任何想法将不胜感激，谢谢！

【参考方案1】：

这似乎是一个简单的解决方案：

library(dplyr)
library(stringr)
data %>%
  mutate(
    # step 1 - paste `color1` and `color2` together and remove " NA":
    color = gsub("\\sNA", "", paste(color1, color2)),
    # step 2 - count the number of white space characters:
    color = str_count(color, " "),
    # step 3 - label `color` as "multicolor" where `color` != 0:
    color = ifelse(color == 0, color1, "multicolor")) %>%
  # remove the obsolete color columns: 
  select(-matches("\\d$"))
  ID      color
1 23        red
2 44 multicolor
3 51     yellow
4 59 multicolor

数据：

data <- data.frame(ID = c(23, 44, 51, 59),
                   color1 = c("red", "blue", "yellow", "green"),
                   color2 = c(NA, "purple", NA, "orange"))

【参考方案2】：

基础 R 方法

# get colors from columns named color*
colo <- paste(names(table(unlist(df1[,grep("color",colnames(df1))]))), collapse="|")

colo
[1] "blue|green|red|yellow|orange|purple"

# match the colors and do the conversion
data.frame( 
  ID=df1$ID, 
  color=apply( df1, 1, function(x) 
    y=x[grep(colo, x)];
    if(length(y)>1)y="multicolor"; y  ) )
  ID      color
1 23        red
2 44 multicolor
3 51     yellow
4 59 multicolor

数据

df1 <- structure(list(ID = c(23L, 44L, 51L, 59L), color1 = c("red", 
"blue", "yellow", "green"), color2 = c(NA, "purple", NA, "orange"
)), class = "data.frame", row.names = c(NA, -4L))

【参考方案3】：

假设data 是您的数据集，您可以这样做。

library(dplyr)

data <- data.frame(ID = c(23, 44, 51, 59),
                   color1 = c("red", "blue", "yellow", "green"),
                   color2 = c(NA, "purple", NA, "orange"))

data %>% 
  mutate(color = ifelse(is.na(color2), color1, "multicolor")) %>% 
  select(ID, color)

【参考方案4】：

这是 tidyverse 中的一种方法。

library(dplyr)
library(tidyr)

df %>% 
  pivot_longer(cols = starts_with("color"), values_to = "color", values_drop_na  = TRUE) %>% 
  group_by(ID) %>% 
  summarize(n = n(),
            color = toString(color), .groups = "drop") %>% 
  mutate(color = if_else(n > 1, "multicolor", color)) %>% 
  select(-n)

# # A tibble: 4 x 2
#      ID color     
#   <int> <chr>     
# 1    23 red       
# 2    44 multicolor
# 3    51 yellow    
# 4    59 multicolor

我是故意这样做的。请注意，如果您在 summarize() 行之后停止，您将获得实际颜色。

# # A tibble: 4 x 3
#      ID     n color        
#   <int> <int> <chr>        
# 1    23     1 red          
# 2    44     2 blue, purple 
# 3    51     1 yellow       
# 4    59     2 green, orange

如果您有许多颜色列，而不仅仅是 2 个，这将扩展。玩弄它，有很多方法可以调整这样的东西。

---

数据

df <- read.table(textConnection("ID  color1   color2
23   red      NA
44   blue     purple
51   yellow   NA
59   green    orange"), header = TRUE)