求助：如何在大数据框(df)中用dplyr的group_by添加Team拼接列？

Hey there! Let's work through this problem together—since you're dealing with a large dataset, ditching loops is absolutely the right move, and dplyr has all the tools you need to pull this off efficiently.

First, let's start with a concrete example to make this tangible. Let's say we have a sample dataframe like this:

library(dplyr)
library(stringr)

# Sample input data
df <- tibble(
  ID = 1:8,
  Team = c("A", "A", "B", "C", "C", "C", "A", "B")
)

Step 1: Add the grouping column #

Since you mentioned the Team column has no fixed pattern, I'll assume you want to group rows by consecutive identical Team values (adjust this logic if your grouping rule is different!). We can generate a unique group ID using vectorized operations (no loops needed):

df <- df %>%
  mutate(Group = cumsum(Team != lag(Team, default = first(Team))) + 1)

Here's what this does:

• lag(Team, default = first(Team)) shifts the Team column down by one row, using the first value as a default for the first row.
• Team != lag(...) creates a boolean flag where TRUE means the Team value changed from the previous row.
• cumsum(...) converts those flags into a running total, which gives us a unique ID for each consecutive group. Adding 1 just starts our group IDs at 1 instead of 0.

Step 2: Add the concatenated Team column #

Now we can use group_by() to group by our new Group column, then create a concatenated string of Team values for each group. Depending on your needs, you can either concatenate all values (including duplicates) or just unique values:

df_processed <- df %>%
  group_by(Group) %>%
  # Option 1: Concatenate unique Team values (good if groups have duplicates)
  mutate(Team_Combined = str_c(unique(Team), collapse = ", ")) %>%
  # Option 2: Concatenate all Team values (including repeats)
  # mutate(Team_Combined = str_c(Team, collapse = ", ")) %>%
  ungroup()

What the final output looks like #

For our sample data, the processed dataframe will look like this (using Option 1):

Adjusting the grouping logic #

If your grouping rule isn't consecutive Team values (e.g., grouping by ID ranges, or another column's criteria), just modify the mutate(Group = ...) line. For example:

• Group every 3 rows: Group = ceiling(ID / 3)
• Group based on a numeric threshold: Group = cumsum(Score > 100) (assuming a Score column)

All these operations are vectorized, so they'll handle large datasets way faster than any for loop would.

内容的提问来源于stack exchange，提问作者Grigorij Abramov