如何在dplyr调用的函数中返回两个数据框？附部分函数代码

First off, let's clear up a key point about R functions: they can only send back one object at a time. So to return two data frames, we'll wrap them up in a list—that's the standard approach here. Let's fix up your existing code first, then show a more efficient dplyr-style alternative since loops can drag with large datasets.

Step 1: Finish Your Loop-Based Code #

Your current code is tracking occupied dates and overwrite counts—let's complete that logic and add a second data frame (we'll use the original testdf with an extra column to track how many times each row's dates were overwritten).

library(dplyr)

correct_occupancy <- function(testdf) {
  # Initialize first data frame: tracks detailed occupied date info
  occupied_dates_df <- data.frame(
    OccupiedDate = as.POSIXct(character()),
    Source = integer(),
    created = as.POSIXct(character()),
    Overwritten_cnt = integer(),
    testdf_row = integer(),
    stringsAsFactors = FALSE
  )
  
  # Initialize second data frame: original data + overwrite count per row
  updated_testdf <- testdf %>%
    mutate(Overwritten_times = 0, .before = 1) # Add column to track overwrites for each row
  
  ovrd_cnt <- 0
  
  for (i in 1:nrow(testdf)) {
    # Generate date range for current booking (check-in to day before check-out)
    testspan <- seq(as.Date(testdf$check_in[i]), as.Date(testdf$check_out[i]) - 1, by = "days")
    # Convert to a data frame with metadata
    testspan_df <- testspan %>%
      as.data.frame() %>%
      rename(OccupiedDate = ".") %>%
      mutate(
        OccupiedDate = as.POSIXct(OccupiedDate),
        Source = i,
        created = Sys.time(),
        Overwritten_cnt = 0,
        testdf_row = i
      )
    
    # Check each date in the current span
    existing_dates <- occupied_dates_df$OccupiedDate
    for (j in 1:nrow(testspan_df)) {
      current_date <- testspan_df$OccupiedDate[j]
      if (current_date %in% existing_dates) {
        # Increment overwrite count for the existing date
        occupied_dates_df$Overwritten_cnt[occupied_dates_df$OccupiedDate == current_date] <- 
          occupied_dates_df$Overwritten_cnt[occupied_dates_df$OccupiedDate == current_date] + 1
        # Update the overwrite count for the current row in testdf
        updated_testdf$Overwritten_times[i] <- updated_testdf$Overwritten_times[i] + 1
        ovrd_cnt <- ovrd_cnt + 1
      } else {
        # Add new date to the occupied dates frame
        occupied_dates_df <- bind_rows(occupied_dates_df, testspan_df[j, ])
      }
    }
  }
  
  # Return both data frames wrapped in a list
  return(list(
    occupied_dates = occupied_dates_df,
    updated_test_data = updated_testdf
  ))
}

How to Use This Function #

Once you run the function, you'll get a list—extract each data frame using $:

# Example call with your test data
result <- correct_occupancy(your_test_data_frame)

# Get the occupied dates details
occupied_dates <- result$occupied_dates

# Get the updated original data with overwrite counts
updated_test_df <- result$updated_test_data

Step 2: Optimize with Dplyr/Tidyr (No Loops!) #

Loops work, but they're slow for large datasets. Here's a vectorized, dplyr-friendly version that does the same job faster:

library(dplyr)
library(tidyr)
library(purrr)

correct_occupancy_vectorized <- function(testdf) {
  # 1. Generate all occupied dates with row metadata
  all_occupied_dates <- testdf %>%
    mutate(
      # Create date range for each booking
      OccupiedDate = map2(check_in, check_out, ~seq(as.Date(.x), as.Date(.y)-1, by = "days")),
      Source = row_number(), # Track which row the date came from
      created = Sys.time()
    ) %>%
    unnest(OccupiedDate) %>% # Expand to one row per date
    mutate(OccupiedDate = as.POSIXct(OccupiedDate))
  
  # 2. Build the occupied dates data frame with overwrite counts
  occupied_dates_df <- all_occupied_dates %>%
    group_by(OccupiedDate) %>%
    mutate(Overwritten_cnt = n() - 1) %>% # n()-1 because first occurrence has 0 overwrites
    ungroup() %>%
    select(OccupiedDate, Source, created, Overwritten_cnt, testdf_row = Source)
  
  # 3. Build the updated test data frame with row-level overwrite counts
  updated_testdf <- all_occupied_dates %>%
    group_by(Source) %>%
    # Count how many times this row's dates were duplicated (overwritten)
    summarize(Overwritten_times = sum(duplicated(OccupiedDate)), .groups = "drop") %>%
    # Join back to original testdf to retain all columns
    right_join(testdf %>% mutate(Source = row_number()), by = "Source") %>%
    relocate(Overwritten_times, .before = 1) %>%
    select(-Source) # Remove the temporary Source column
  
  # Return both data frames in a list
  list(
    occupied_dates = occupied_dates_df,
    updated_test_data = updated_testdf
  )
}

This version uses map2 to generate date ranges, unnest to expand into long format, and grouping to calculate counts—all operations optimized for speed in tidyverse packages.

内容的提问来源于stack exchange，提问作者Snehal Kakade