如何将多组x、y列的DataFrame转换为单组x、y列的DataFrame？

Alright, let's solve this DataFrame reshaping task using tidyr::gather and tidyr::spread as you asked—no loops needed! Here's a step-by-step breakdown with example code you can follow directly.

Step 1: Prepare Example Data (for testing) #

First, let's create a sample a DataFrame that matches your column structure, so you can replicate the process:

library(tidyr)
library(dplyr) # For helper functions like `select` and pipe operator

# Sample DataFrame
a <- data.frame(
  x1 = c(1, 2, 3),
  y1 = c(4, 5, 6),
  x2 = c(7, 8, 9),
  y2 = c(10, 11, 12),
  x3 = c(13, 14, 15),
  y3 = c(16, 17, 18)
)

Step 2: Reshape Step-by-Step #

1. Convert Wide to Long Format with gather #

First, we use gather to collapse all columns into two: one for the column names (e.g., x1, y1) and one for their corresponding values:

a_long <- gather(a, key = "variable", value = "value")

This gives us a long DataFrame where each row represents a single value from the original columns, paired with its column name.

2. Split Column Names into Type and Group #

Next, we split the variable column into two parts: the type (x or y) and the group number (1, 2, 3). We use tidyr::separate for this:

a_separated <- separate(a_long, col = variable, into = c("type", "group"), sep = 1)

The sep = 1 argument tells R to split the string after the first character (since our column names start with x/y followed by a number).

3. Reshape Back to Wide Format (x & y Columns) #

Now we use spread to turn the type column into two separate columns (x and y), using the value column for their data:

a_reshaped <- spread(a_separated, key = type, value = value)

4. Clean Up (Optional) #

If you don't need the group column anymore, you can drop it with dplyr::select:

a_final <- select(a_reshaped, -group)

Combine All Steps with Pipes #

You can also chain all these operations together using the pipe operator (%>%) for cleaner code:

a_final <- a %>%
  gather(key = "variable", value = "value") %>%
  separate(variable, into = c("type", "group"), sep = 1) %>%
  spread(key = type, value = value) %>%
  select(-group)

Result #

The final a_final DataFrame will have just two columns: x and y, with all the values from the original x1/x2/x3 and y1/y2/y3 columns stacked appropriately.

内容的提问来源于stack exchange，提问作者Minyi Han