在R中拆分含id列的数据框为所有3列组合（data.table适用）

Let's fix your approach and make it efficient for your large dataset. The key issue in your original code was selecting 3 non-ID columns (then adding ID, resulting in 4-column tables) instead of selecting 2 non-ID columns to pair with ID for 3-column tables total. Here's the optimized solution using data.table:

Step-by-Step Solution #

1. Convert your data frame to data.table (critical for handling 1M+ rows efficiently):

   library(data.table)
   dt <- as.data.table(frame)
   

2. Identify non-ID columns:

   non_id_cols <- setdiff(names(dt), "id")
   

3. Generate all 2-column combinations from non-ID columns:
   Using combn(..., simplify = FALSE) gives us a list of character vectors, each representing a pair of columns to pair with ID:

   column_combos <- combn(non_id_cols, 2, simplify = FALSE)
   

4. Create list of 3-column data.tables:
   Use lapply to iterate over each combination and subset the data.table. This is efficient because data.table avoids unnecessary data copies (it references existing columns):

   list_tables <- lapply(column_combos, function(col_pair) {
     dt[, c("id", col_pair)]  # Select ID + the two columns from the combo
   })
   

Key Improvements Over Your Original Code #

• Correct column count: We're pairing ID with 2 non-ID columns to get exactly 3 columns per table, matching your requirement.
• Efficiency: Using data.table column selection and lapply avoids messy transposing/rbinding that would slow down your large dataset.
• Readability: The code is straightforward and easy to modify if you need to adjust the number of columns later.

Optional: Name the List Elements #

To make your list easier to work with, you can name each element after the columns it contains:

names(list_tables) <- sapply(column_combos, function(cols) {
  paste0("id_", paste(cols, collapse = "_"))
})

Saving the List for Later Use #

To save the list for future operations, use save():

save(list_tables, file = "3col_id_combinations.RData")

When you need to load it later:

load("3col_id_combinations.RData")

内容的提问来源于stack exchange，提问作者iomedee