One of the most important metrics for airlines is their on-time flight performance. The Federal Aviation Administration considers a flight delayed when it arrives at least 15 minutes later than its scheduled arrival time. Pandas has direct methods to calculate the total and percentage of on-time flights per airline. While these basic summary statistics are an important metric, there are other non-trivial calculations that are interesting, such as finding the length of consecutive on-time flights for each airline at each of its origin airports.
In this recipe, we find the longest consecutive streak of on-time flights for each airline at each origin airport. This requires each value in a column to be aware of the value immediately following it. We make clever use of the diff and cumsum methods in order to find streaks before applying this methodology to each of the groups.