如何批量为DataFrame的cc列生成累积和？技术求助

Hey there! Let's fix this cumulative sum issue for your cc columns. The key here is that you need to compute a running total for each cc column, and since you have an unknown number of these columns later on, we need a scalable approach that doesn't require hardcoding each column name.

First, let's break down why your previous apply attempts didn't work:

• In your functions, you tried to modify the original data frame (df$x) directly instead of returning the processed column. When using apply on a data frame, each column is passed to your function as a vector, and your function should return the modified vector—apply will then reassemble these vectors into a new data frame.
• You also referenced df$x inside the function, which doesn't correspond to the current column being processed by apply; you should use the input parameter x instead.

Let's start with a reproducible example of your data: #

df1 <- data.frame(
  rr_1 = c(100, 200, 300, 400, 0),
  rr_2 = c(0, 100, 300, 500, 0),
  cc_1 = c(1, 1, 1, 1, 0),
  cc_2 = c(0, 1, 1, 1, 0)
)

Method 1: Using dplyr (recommended for tidy data workflows) #

If you're using dplyr (version 1.0.0 or newer), across() is perfect for batch-processing columns. We can target all columns starting with "cc", apply cumsum() to each, and update them in place:

library(dplyr)

df_processed <- df1 %>%
  mutate(across(starts_with("cc"), cumsum))

Let's check the result:

print(df_processed)
#   rr_1 rr_2 cc_1 cc_2
# 1  100    0    1    0
# 2  200  100    2    1
# 3  300  300    3    2
# 4  400  500    4    3
# 5    0    0    4    3

This matches exactly what you're looking for! The cumsum() function naturally handles the final 0 values by carrying forward the last cumulative total, which is exactly your desired behavior.

Method 2: Using base R apply #

If you prefer base R, we can adjust your approach to work correctly. The key is to have the function return the processed column instead of modifying the original data frame:

# Extract cc columns
cc_cols <- df1 %>% select(starts_with("cc"))

# Apply cumsum to each column; apply returns a matrix, so convert back to data frame
cc_processed <- as.data.frame(apply(cc_cols, 2, cumsum))

# Combine with the non-cc columns
df_processed_base <- cbind(df1 %>% select(-starts_with("cc")), cc_processed)

This will give you the same result as the dplyr method.

Why this works #

Your target cumulative sequence is exactly what cumsum() produces:

• For cc_1, the original values are [1,1,1,1,0] → cumsum() gives [1,2,3,4,4]
• For cc_2, original values are [0,1,1,1,0] → cumsum() gives [0,1,2,3,3]

No need for custom loops—R's built-in cumsum() does exactly what you need, and using across() or apply lets you apply it to all cc columns automatically, regardless of how many there are.

内容的提问来源于stack exchange，提问作者Luis Manso