使用Reduce函数合并R数据框时获取数据框名称并重命名重复列

Great question! I’ve run into this exact frustration before—those default .x/.y suffixes from merge() get really hard to track when combining multiple data frames. Let’s walk through how to fix this with Reduce() while keeping your column names clear and tied to the original data frame names.

The Core Problem #

When you use Reduce(merge, df_list), merge() automatically appends .x/.y to duplicate column names (excluding your by column). The issue is that you can’t easily map these generic suffixes back to your original data frame names, since names() only gives you column names, not the names of the data frames themselves.

The Solution: Rename Columns Before Merging #

Instead of letting merge() generate generic suffixes, we’ll pre-rename duplicate columns with their parent data frame’s name first. This way, when we merge, there are no duplicate column names to begin with, and we avoid the .x/.y mess entirely.

Step 1: Create a Helper Function to Rename Columns

First, write a small function that renames all non-by columns in a data frame to include the data frame’s name as a suffix:

rename_non_by_cols <- function(df, by_col, df_name) {
  # Identify columns that aren't the merge key
  non_by_cols <- setdiff(names(df), by_col)
  # Rename those columns with the data frame's name as a suffix
  names(df)[names(df) %in% non_by_cols] <- paste0(non_by_cols, "_", df_name)
  return(df)
}

Step 2: Use Reduce() with Pre-Renamed Data Frames

Let’s use a concrete example to see this in action. Suppose we have three data frames with overlapping id columns and a shared value column:

# Sample data frames
df1 <- data.frame(id = 1:3, value = c("a", "b", "c"))
df2 <- data.frame(id = 2:4, value = c("d", "e", "f"))
df3 <- data.frame(id = 3:5, value = c("g", "h", "i"))

# Store them in a named list (critical—we need the data frame names!)
df_list <- list(df1 = df1, df2 = df2, df3 = df3)

Now, we’ll first rename the columns in each data frame, then use Reduce() to merge them all together:

# Define our merge key
by_column <- "id"

# Rename columns in each data frame using our helper function
renamed_dfs <- lapply(names(df_list), function(name) {
  rename_non_by_cols(df_list[[name]], by_column, name)
})

# Merge all renamed data frames with Reduce()
final_merged_df <- Reduce(function(x, y) merge(x, y, by = by_column, all = TRUE), renamed_dfs)

If you print final_merged_df, you’ll see clean column names: id, value_df1, value_df2, value_df3—no more generic suffixes!

A More Concise Version with purrr (Optional) #

If you’re comfortable with the purrr package, you can streamline this even further using imap() (which lets you access both the data frame and its name) and reduce():

library(purrr)

final_merged_df <- df_list %>%
  imap(function(df, df_name) rename_non_by_cols(df, "id", df_name)) %>%
  reduce(~merge(.x, .y, by = "id", all = TRUE))

Why This Works Better Than the Default Reduce(merge, ...) #

By pre-renaming columns, we eliminate duplicate column names before merging. This means merge() doesn’t need to generate .x/.y suffixes, and every column’s name clearly tells you which data frame it came from. It’s essentially doing the same work as your for loop, but wrapped in a clean, functional programming style.

内容的提问来源于stack exchange，提问作者Basti