在R中统计每架飞机首次1小时以上延误前的航班数量

Great question! Let's work through this step by step to get the count of flights each plane made before its first 60+ minute arrival delay.

First, let's fix a couple of issues with your current approach:

• Your grouping includes too many columns (like arr_delay), which splits a single plane's flights into unnecessary subgroups.
• Using first() without properly sorting flights by time won't give you the chronological first delay, which is critical here.

Step-by-Step Code & Explanation #

Here's a tidyverse-based solution that correctly orders flights, tracks the first delay, and counts the preceding flights:

library(tidyverse)

flight_delay_summary <- flights %>%
  # Keep only relevant columns to simplify processing
  select(tailnum, year, month, day, dep_time, arr_delay) %>%
  # **Critical**: Sort flights by plane, then by date/time to ensure chronological order
  arrange(tailnum, year, month, day, dep_time) %>%
  group_by(tailnum) %>%
  # Mark which flights had a 60+ minute delay
  mutate(is_long_delay = arr_delay > 60) %>%
  # Track whether we've already seen the first long delay (cumsum turns TRUE once a delay occurs)
  mutate(has_seen_first_delay = cumsum(is_long_delay) >= 1) %>%
  # Filter to keep only flights up to and including the first long delay
  filter(!has_seen_first_delay | (has_seen_first_delay & is_long_delay)) %>%
  # Calculate the number of flights *before* the first long delay
  summarise(
    flights_before_first_long_delay = n() - ifelse(any(is_long_delay), 1, 0),
    # Optional: Add a flag if the plane ever had a long delay
    ever_had_long_delay = any(is_long_delay)
  )

How This Works: #

1. Sorting: We first sort each plane's flights by date and departure time to make sure we're checking delays in the order they actually happened.
2. Delay Flag: is_long_delay simply marks each flight as meeting the 60+ minute delay condition.
3. Track First Delay: has_seen_first_delay uses cumsum()—once a TRUE (delay) is encountered, all subsequent rows become TRUE, letting us know we've passed the first delay point.
4. Filter Rows: We keep all flights before the first delay, plus the first delay flight itself.
5. Count Preceding Flights: For planes that had a delay, we subtract 1 from the total row count (to exclude the delay flight itself). For planes that never had a delay, we just use the total number of flights.

Example Check #

Looking at your sample data:

• For D942DN, the first flight was already a 91-minute delay. So flights_before_first_long_delay will be 0.
• For N0EGMQ, the second flight was the first delay. So flights_before_first_long_delay will be 1.

This will give you a clean summary for all 4,044 unique tailnumbers.

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