如何优化R语言中把左侧有非NA值的非NA值置为NA的代码？

Let's solve this problem where we need to retain only the first non-NA value in each consecutive block of non-NA elements, and set all other non-NA values in those blocks to NA. Here are a few concise, optimized approaches:

1. Base R (No External Packages) #

This method uses base R's built-in rle() (run-length encoding) to identify consecutive blocks, then creates an index to keep only the first non-NA in each block:

a <- c(3,2,3,NA,NA,1,NA,NA,2,1,4,NA)

# Get run-length encoding of non-NA status
r <- rle(!is.na(a))
# For non-NA runs, mark only the first element as keepable
r$values <- mapply(function(val, len) if (val) c(TRUE, rep(FALSE, len-1)) else val, r$values, r$lengths)
# Convert back to logical index matching the original vector
keep_idx <- inverse.rle(r)
# Set non-keep positions to NA
a[!keep_idx] <- NA

a
# Output: [1] 3 NA NA NA NA 1 NA NA 2 NA NA NA

How it works: rle() breaks down the vector into consecutive runs of NA/non-NA. We modify the run values to only keep the first element of each non-NA run, then use inverse.rle() to map this back to the original vector length.

2. Optimized data.table Approach #

If you're already using data.table, we can condense your original solution into a single line—no need for intermediate variables:

library(data.table)
a <- c(3,2,3,NA,NA,1,NA,NA,2,1,4,NA)

a[duplicated(rleidv(!is.na(a)) & !is.na(a))] <- NA

a
# Output: [1] 3 NA NA NA NA 1 NA NA 2 NA NA NA

How it works: rleidv() generates a unique ID for each consecutive block of NA/non-NA. We combine this with !is.na(a) to target only non-NA elements, then use duplicated() to flag all elements after the first in each non-NA block.

3. tidyverse/dplyr Approach #

For users who prefer the tidyverse workflow, we can group consecutive non-NA blocks and keep only the first non-NA per group:

library(dplyr)
a <- c(3,2,3,NA,NA,1,NA,NA,2,1,4,NA)

result <- tibble(val = a) %>%
  # Create groups for consecutive non-NA blocks
  group_by(block = cumsum(is.na(lag(val, default = TRUE)))) %>%
  # Keep only first non-NA in each block, set others to NA
  mutate(val = ifelse(row_number() > 1 & !is.na(val), NA, val)) %>%
  ungroup() %>%
  pull(val)

result
# Output: [1] 3 NA NA NA NA 1 NA NA 2 NA NA NA

How it works: cumsum(is.na(lag(val))) creates a unique group ID every time an NA is encountered (marking the start of a new block). We then use row_number() to target elements after the first in each non-NA group.

All these methods will give you the desired output, with options to fit your preferred coding style or package ecosystem.

内容的提问来源于stack exchange，提问作者Andre Elrico