New model more accurately predicts our daily travel choices

EPFL engineers developed a predictive model that takes into account not only your commuting habits, but also your daytime activities. Her flexible approach incorporates the idea of ​​trade-offs to provide more realistic predictions.

Transportation engineers often use computer models to estimate demand for specific itineraries, such as how many cars travel on the Lausanne-Geneva motorway each year and which rail lines carry the most passengers. Answer questions such as. This is a broad and fascinating field, and one that Janodi Pougala, a student in the Department of Civil Engineering at EPFL's Institute of Transport and Mobility, decided to study for his PhD. paper.

Pougala has developed a new model for predicting individuals' travel choices. This model takes into account a wider range of variables and more closely maps actual behavior. Her program, which is available in open source, focuses not only on how people normally move, but also on their daily activities. This represents a particularly sophisticated approach to explaining how people react to the unpredictable events that inevitably form part of our daily lives.

In the traditional model, transportation engineers begin by studying each trip an individual takes, along with the reason for the trip, the mode of transportation the user uses, and the itinerary chosen. The engineer then develops a program that describes this behavior in chronological order. However, these programs are often poorly suited to complex realities.

Modeling trade-offs

To design more accurate models, engineers need a deeper understanding of how people behave. That's especially true considering today's increasingly diverse lifestyles. Commuting patterns have changed significantly as more people work from home, car-sharing systems are introduced, and infrastructure improvements that allow employees to live further away from their employers. These are some of the structural changes that Pugala wanted to address with the new model, which is considered more accurate because it is based on individual activities and preferences.

How does the model work? “We start by scheduling an individual's daily activities, and then we connect the corresponding variables in a formula,” Poogala says. “We extracted data for the variables from a variety of sources, including commuting surveys and statistical results.” The key to her model is its highly flexible design. “Instead of analyzing the factors in sequence, we analyze all the factors at the same time,” she says.

Additionally, her model is not bound to a predefined order of events during the day, allowing for decisions based on individual satisfaction and constraints. So this is a new way to model tradeoffs. Pougalla took behavioral hypotheses described in sociology and urban environment literature and research and translated them into mathematical formulas.

She then combined the equations with statistical data to ensure the model made the most realistic predictions possible. For example, let's say a woman named Emma stays late at her job and decides not to go to the gym. On her way home, her train encountered technical problems at Lausanne station. Instead of waiting for an alternative train, Emma decides to take a bus.

Pougala says, “My model can predict how different individuals will react under these circumstances, and how long they can tolerate less favorable situations. It can also predict how people We can also explain how to adapt and use alternative transportation methods.”

City officials can use Puugara's model in long-term planning to decide what type of transportation infrastructure to develop. It has already been tested in a model used by Swiss railway companies and in an urban planning project in Zurich that aims to show what the city would look like if half of the transport there was not electric. Masu.