如何用dplyr按MatchID分组计算胜负球队KPI差值？

Absolutely, you can solve this with dplyr—let’s fix that error and get you the win-loss KPI differences you need.

First, let’s unpack the error you got: Error in inherits(x, "fun_list") : object 'Outcome' not found. This usually happens when the function you pass to summarise_at can’t access the Outcome column correctly in the grouped context, especially since summarise_at is the older dplyr syntax. The modern approach using across() is more intuitive and avoids this kind of context issue.

Step 1: Validate Your Data #

First, make sure every MatchID has exactly one "Win" and one "Loss" entry—this is critical for the calculation to work reliably:

library(dplyr)

# Check for matches with missing or duplicate outcomes
df %>%
  count(MatchID, Outcome) %>%
  filter(n != 1)

If this returns any rows, you’ll need to clean your data first (e.g., remove incomplete matches or resolve duplicates).

Step 2: Calculate Win-Loss KPI Differences #

Use group_by() to group by each match, then across() to compute the difference for every KPI column. Here’s the code:

# Calculate Win - Loss for each KPI per MatchID
match_kpi_diffs <- df %>%
  group_by(MatchID) %>%
  summarise(
    # Target all your KPI columns (adjust the selector to match your data)
    across(
      starts_with("Var"),  # Use this if KPIs are named Var1, Var2, etc.
      # Or specify exact columns: all_of(c("Var1", "Var2", "Var3", "Var4", "Var5"))
      ~ .[Outcome == "Win"] - .[Outcome == "Loss"]
    ),
    .groups = "drop"  # Ungroup after summarizing
  )

# Preview the result
head(match_kpi_diffs)

How This Works #

• group_by(MatchID) ensures we’re only working with one match at a time.
• across() lets us apply the same calculation to multiple columns in one go. The selector starts_with("Var") grabs all your KPI columns—adjust this if your KPIs have different names (e.g., contains("KPI") or explicit column names with all_of()).
• For each KPI column, .[Outcome == "Win"] picks the value from the winning team, and we subtract the losing team’s value (.[Outcome == "Loss"]). Since each group has exactly one win and one loss, this gives a single difference value per match and KPI.

Why Your Original summarise_at Failed #

Older summarise_at syntax requires wrapping functions in a list, and the context for accessing Outcome can be tricky. For example, a correct (but outdated) version would look like this, but we recommend using across() instead for readability and compatibility:

# Outdated summarise_at approach (not recommended)
df %>%
  group_by(MatchID) %>%
  summarise_at(
    vars(starts_with("Var")),
    funs(.[Outcome == "Win"] - .[Outcome == "Loss"])
  )

This should give you the exact dataset you want: one row per MatchID, with each column being the winning team’s KPI minus the losing team’s corresponding KPI.

内容的提问来源于stack exchange，提问作者user2716568