R语言：如何排除指定变量组全为NA的数据行？

Hey there! Let's sort out that error you ran into when trying to filter your dataframe. The problem with your original code is two-fold: missing a critical comma in the row index, and checking for the wrong condition (you want to keep rows that aren't all NA in columns 4-22, not the ones that are).

Base R Solution #

Here's the corrected base R code that does exactly what you need:

# Keep rows where columns 4-22 are NOT all NA
df <- df[rowSums(is.na(df[, 4:22])) != 19, ]

Let me break this down:

• df[, 4:22] targets the columns you care about (4 through 22, which is 19 total columns)
• is.na() converts all NA values in those columns to TRUE
• rowSums() counts how many NAs are in each row of those columns
• We keep rows where that count isn't 19 (meaning at least one value is non-NA)
• The trailing comma [, ] ensures we keep all columns (including your first 3, regardless of their NA status)

Tidyverse (dplyr) Solution #

If you prefer a more readable approach using the tidyverse, this works too—no need to calculate the number of columns:

library(dplyr)

df <- df %>%
  filter(if_any(4:22, ~ !is.na(.x)))

This code directly checks: "Does this row have at least one non-NA value in columns 4-22?" If yes, it stays. Super straightforward!

Either of these methods will keep your first 3 columns intact, ignore their NA status, and only remove rows where columns 4-22 are completely filled with NAs.

内容的提问来源于stack exchange，提问作者J. Doe