But the awkward pandemic years have made all this more complicated. Oren Etzioni is now the CEO of the Allen Institute for AI, but in the early 2000s, he built and sold one of the first airfare forecasting tools. Prediction algorithms are pretty good at reweighing the importance of different factors as the world changes, and, he says, “they have a chance to adapt automatically by having the most recent data available.” But that could take a while, according to Etzioni: days, if not weeks.
Google Flights helps customers find the cheapest tickets for their favorite routes and dates. But since the spring of 2020, the search engine has significantly reduced the number of “predictive insights” – predictions of when prices are likely to rise or fall. Overall, Flights aims for a 90 percent forecast accuracy, said Eric Zimmerman, director of travel products at Google. “With increased volatility in airline tickets, it has become more difficult to achieve that high level of confidence,” he says. The pandemic and its effects on air traffic also prompted the company to halt an experiment launched in the summer of 2019 where it would guarantee fares for certain specific routes and send pilots refunds if the pre-takeoff price fell. valleys. It could quickly bring the project back, Zimmerman says, as the industry begins to stabilize.
Giorgos Zacharia, president of online travel agency and search engine Kayak, says he has a team of MIT doctoral students who spend their working lives working on the website’s price prediction tool. While the prediction algorithm, which first launched in 2013, usually needs tweaking every few years, he says, the past two have undergone “serious retraining” every few months and sometimes every few weeks. He says the accuracy of the forecasting tools, which is generally around 85 percent, may have dropped periodically in recent years — perhaps closer to 83 percent. That means that waiting at some lows or buying when the website said so was less likely to have led to the lowest possible price – and could have led to some light fistfights to the skies instead.
“Machine learning likes to learn from old and repeatable patterns of the past and make predictions based on the likelihood of those patterns working again,” Zachariah says. “So the pandemic, which brings a lot of unexpected outliers, is also impacting the input data from models like this one and making it a more challenging environment.”
Hayley Berg, the chief economist at Hopper, says the company’s predictive tool has been trained on 75 trillion routes and eight years of historical price data. But today, the algorithm outweighs what it’s seen over the past three years, keeping the tool at 95 percent accuracy throughout the pandemic, according to the company. Even in the first few days of Covid-related shutdowns, she says, Hopper had accurate airfare price predictions 90 percent of the time. Still, customers shouldn’t be alarmed by price volatility – Hopper has found that the average domestic flight changes price 17 times in two days, and 12 times if it’s an international flight.
All those changes are sparking plenty of conspiracy theories among ticket buyers, even those not involved with price prediction platforms. No, executives say, airlines don’t track cookies and don’t charge prices if they see you’re interested in a particular route. (Zacharia, the Kayak president, does say that sometimes the fares are higher or lower depending on your location when you search, because the systems take into account “points of sale”.) No, there’s no reason flights should be cheaper on a Tuesday than any other day, a persistent rumor among bargain hunters. “The best time to book depends on your trip, specifically the origin, destination, departure, and return,” Berg says. “And it can be vastly different depending on where you’re going.”
Today, however, it doesn’t always take an advanced machine learning algorithm to pick the best time to buy – there’s no right time. Prices are so high, says Victoria Hart, a spokesperson for Kayak, that there aren’t many “wait” indicators these days.
