R dplyr 使用 group_by 总结均值和标准差

Question

我有一个如下所示的数据框：

df <- data.frame("Experiment" = c(rep("Exp1", 6), rep("Exp2", 5), rep("Exp3", 4)),
                 "Replicate" = c("A","A","A","B","C","C","A","A","B","B","C","A","B","B","C"),
                 "Type" = c("alpha","beta","gamma","alpha","alpha","beta","alpha","gamma","beta","gamma","beta","alpha","alpha","gamma","beta"),
                 "Frequency" = c(10,100,1000,15,5,105,10,1010,95,1020,105,15,10,990,100))

我正在尝试为 Experiment 和 Type 的组合计算 Frequency 的均值和标准差，我首先通过运行宁此行进行了尝试：

df %>% group_by(Experiment, Type) %>% summarise(mean = mean(Frequency), sd = sd(Frequency)

如果我运行这样做，我会得到如下所示的提示：

Experiment   Type   mean   sd
Exp1         alpha  10     5
Exp1         beta   102.   3.54
Epx1         gamma  1000   NA

但我希望 R 认为所有 Type（alpha、beta、gamma）对于 Experiment 的每个组合都应该存在和 Replicate，这样如果 Type 没有 Frequency 值，R 将使用 0 而不是不包括该值。

换句话说，我想要的需要像下面这样计算：

Experiment   Type   mean              sd
Exp1         alpha  mean(10,15,5)     sd(10,15,5)
Exp1         beta   mean(100,0,105)   sd(100,0,105)
Exp1         gamma  mean(1000,0,0)    sd(1000,0,0)

例如，对于Exp1beta，我上面使用的summarise函数计算mean(100,105)和sd(100,105)，因为Exp1Replicate B 在我的 df 中不存在。但我希望 R 改为计算 mean(100,0,105) 和 sd(100,0,105)。谁能给我一些关于如何做到这一点的想法？

Answer 1

您需要先 complete 您的数据框用 0 填充缺失的数据，然后将“已完成”的数据框通过管道传输到您的函数。

library(tidyverse)

df %>% 
  complete(Experiment, Type, Replicate, fill = list(Frequency = 0)) %>% 
  group_by(Experiment, Type) %>% 
  summarise(mean = mean(Frequency), sd = sd(Frequency), .groups = "drop")

# A tibble: 9 × 4
  Experiment Type    mean     sd
  <chr>      <chr>  <dbl>  <dbl>
1 Exp1       alpha  10      5   
2 Exp1       beta   68.3   59.2 
3 Exp1       gamma 333.   577.  
4 Exp2       alpha   3.33   5.77
5 Exp2       beta   66.7   58.0 
6 Exp2       gamma 677.   586.  
7 Exp3       alpha   8.33   7.64
8 Exp3       beta   33.3   57.7 
9 Exp3       gamma 330    572.

Answer 2

您需要在 group_by 函数中包含 Replicate 并将输出转换为更宽的小标题。可以通过替换 NA 值来改变数字列。然后，连接 mean 和 sd 列将给出所需的输出。

df %>% group_by(Experiment, Type, Replicate) %>% 
  summarise(mean = mean(Frequency), sd = sd(Frequency)) %>% 
  pivot_wider(names_from = Replicate, values_from =  c(mean, sd)) %>% 
mutate(across(where(is.double),~ replace_na(.,0))) %>% 
  mutate(mean = paste0("mean(", mean_A, ",", mean_B, ",", mean_C, ")"),
         sd = paste0("sd(", sd_A, ",", sd_B, ",", sd_C, ")")) %>% 
  select(Experiment, Type, mean, sd)

输出为

# A tibble: 9 x 4
# Groups:   Experiment, Type [9]
  Experiment Type  mean              sd       
  <chr>      <chr> <chr>             <chr>    
1 Exp1       alpha mean(10,15,5)     sd(0,0,0)
2 Exp1       beta  mean(100,0,105)   sd(0,0,0)
3 Exp1       gamma mean(1000,0,0)    sd(0,0,0)
4 Exp2       alpha mean(10,0,0)      sd(0,0,0)
5 Exp2       beta  mean(0,95,105)    sd(0,0,0)
6 Exp2       gamma mean(1010,1020,0) sd(0,0,0)
7 Exp3       alpha mean(15,10,0)     sd(0,0,0)
8 Exp3       beta  mean(0,0,100)     sd(0,0,0)
9 Exp3       gamma mean(0,990,0)     sd(0,0,0)

R dplyr 使用 group_by 总结均值和标准差

R dplyr summarise mean and stdev using group_by

r

dplyr

summarize