• Digging into data shows which flights are truly late

    This is just part of the equation (fivethirtyeight.com image)

    This is just part of the on-time equation. (fivethirtyeight.com image)

    Patrick Coleman
    I don’t remember when I finally figured out that airlines’ departure and arrival times were significantly padded, but I am pretty sure it was on a flight from Chicago to Minneapolis.

    The amount of time a flight is actually in the air, wheels-up to wheels-down, between O’Hare and Minneapolis, is a little over an hour, maybe 75 minutes. But the posted flight time is anywhere from an hour and a half to an hour and 45 minutes.

    This is done, in part, to improve airlines’ on-time ratings, which is what the U.S. government measures and allows flights to be up to 15 minutes late and still be considered on time. (Never mind if you have 34 minutes to change planes at O’Hare.)

    But Nate Silver of fivethirtyeight.com has made a living digging past the spin and the surface analysis and uncovering the truth in actual data. He’s taken the same kind of rigor that helped him call the 2012 U.S. presidential election in detail and applied it to the big data problem of airlines’ on-time performance.

    Some of the findings are not surprising, such as the fact that Chicago O’Hare and the three major New York City-area airports are big sources of delayed flights. But where the analysis truly shines is in its detail. Which airport pairings are the slowest? Which airline is statistically the fastest to get from Point A to Point B?

    If you have a choice as to which airline to fly on a particular route, it’s worth investigating the data to see how each airline performs. If you are stuck flying a particular airline on a particular route, the data can tell you how likely you are to truly be on time.

    Because there are times when “within 15 minutes” is not on-time enough.