R语言：无需循环，基于同ID条件替换数据框值

Hey there! I totally get the pain of waiting hours for a loop to finish on large datasets—let's swap that slow for loop out with fast, vectorized operations that'll handle your 100k-row frame in seconds instead of hours.

First, Let's Simplify the Condition #

Looking at your rules, we can rephrase them to make things easier:
For each column X (A-E), we want X.1 to take the value of acumulativediff if:

• The current row shares the same id as the previous row, and
• The previous row's X value is 1 (both your original conditions boil down to this—whether current X is 0 or 1, we only care if the prior row's X was 1 and same id).

This simplification lets us use grouping and lag functions to avoid looping entirely.

Solution 1: Using dplyr (Clean & Fast) #

dplyr is optimized for big data operations, so this will be lightning fast. We'll group by id, then use lag() to check the prior row's value for each X column, and update X.1 accordingly.

library(dplyr)
library(stringr) # For string manipulation in the batch version

# Option 1: Explicit column updates (easy to read)
df_processed <- df %>%
  group_by(id) %>%
  mutate(
    A.1 = ifelse(lag(A, default = 0) == 1, acumulativediff, 0),
    B.1 = ifelse(lag(B, default = 0) == 1, acumulativediff, 0),
    C.1 = ifelse(lag(C, default = 0) == 1, acumulativediff, 0),
    D.1 = ifelse(lag(D, default = 0) == 1, acumulativediff, 0),
    E.1 = ifelse(lag(E, default = 0) == 1, acumulativediff, 0)
  ) %>%
  ungroup()

# Option 2: Batch processing (great if you have 100 columns instead of 5!)
original_cols <- colnames(df)[3:7] # A-E
target_cols <- colnames(df)[8:12] # A.1-E.1

df_processed <- df %>%
  group_by(id) %>%
  mutate(
    across(
      .cols = all_of(target_cols),
      .fns = ~ ifelse(lag(get(str_remove(cur_column(), "\\.1")), default = 0) == 1, acumulativediff, 0),
      .names = "{.col}"
    )
  ) %>%
  ungroup()

Solution 2: Base R (No External Libraries) #

If you prefer sticking to base R, we can use ave() to handle grouping and vectorized operations:

original_cols <- colnames(df)[3:7]
target_cols <- colnames(df)[8:12]

for (i in seq_along(original_cols)) {
  orig_col <- original_cols[i]
  targ_col <- target_cols[i]
  
  df[[targ_col]] <- ave(
    df[[orig_col]],
    df$id,
    FUN = function(x) {
      # Create lagged version of the column (first row gets 0)
      lag_x <- c(0, x[-length(x)])
      # Apply the condition
      ifelse(lag_x == 1, df$acumulativediff, 0)
    }
  )
}

Why This Is Way Faster #

Both solutions avoid row-by-row loops:

• dplyr uses optimized C++ under the hood for grouping and transformations.
• Base R's ave() applies vectorized functions per group, which is orders of magnitude faster than iterating over each row.

You can verify the result matches your target with all.equal(df_processed, your_target_df)—it should return TRUE.

内容的提问来源于stack exchange，提问作者torakxkz