R 中的条件统计（首选 dplyr 解决方案）

Question

I have the following data frame:

df <- data.frame("num1" = 1:3, "num2" = 4:6, "num3" = c(NA, 10, 12), stringsAsFactors = FALSE)

  num1 num2 num3
    1    4   NA
    2    5   10
    3    6   12

Is there a way to generate a summary table using the mean for every column conditionally? To elaborate, if a column in the data frame contains a null value, then na.exclude that null and compute the mean: (10 + 12) /2 = 11. If a column does not have any null values, then just compute the mean: eg (1 + 2 + 3) / 3 = 2 for num1 column.

Desired output:

  mean_num1   mean_num2   mean_num3
     2             5          11

Answer 1

You could loop over all columns with an "apply" function, with the mean() function, with the "na.rm=TRUE" argument.

Something like

sapply(df, mean, na.rm=TRUE)

num1 num2 num3 
   2    5   11 

Then you could rename the vector as you please:

names(mean_col)<-paste0('mean_', names(mean_col))

mean_num1 mean_num2 mean_num3 
        2         5        11 

With dplyr:

df%>%summarize(across(everything(), mean, na.rm=TRUE, .names = "mean_{col}"))

Edit

Or the simplest of all, with colMeans() :

colMeans(df, na.rm=TRUE)