Building better prediction models for consumer choices
As consumers, we are constantly making choices about things like what to eat at a restaurant, what clothing to buy at a store and the type of car we'd like to purchase, and so much more. Savvy producers of such goods look to economists to help estimate what products people want and how demand might shift when factors like price and availability change.
Researchers at the California Institute of Technology (Caltech) have made significant progress in improving prediction models for consumer choices by addressing limitations in the random utility model (RUM). This fundamental framework for rational decision-making is widely used in economics, marketing, and other fields to estimate consumer preferences and demand shifts.
However, the model struggles when the full variety of options is unknown, leading to the linear ordering problem, which has remained unsolved since the 1980s.
Now, an interdisciplinary team led by Professor Kota Saito and graduate student Alec Sandroni has successfully solved this challenge. Their findings, published in the August issue of the American Economic Review, demonstrate that even without observing all the unobserved options, economists can extract valuable information from the alternatives that weren't chosen. This discovery could lead to more accurate predictions of individual consumer choices when all options aren't clearly defined.
The researchers applied network flow theory—a computational method used to model the movement of entities like traffic and fluids—to clarify the limitations of the outside-option approach, which groups unobserved alternatives into a single "outside option." They found that using this innovative method, they could test RUM with less data than previously thought necessary, potentially leading to more accurate demand evaluations across various industries.
Alec Sandroni, who discovered his passion for economics through a challenging math test given by Professor Saito during his undergraduate career, played a crucial role in this breakthrough. Despite initially having no background in economics, Sandroni quickly learned the necessary skills and contributed to generating ideas, completing rigorous proofs, and finalizing the manuscript. His dedication and ability to work at a high level as an undergraduate student were instrumental in this achievement.
The team's research also highlights the ambiguity that arises when aggregated choices are represented, as the grouped labels may hide different underlying options for different people or markets. This ambiguity weakens the testable implications of RUM. Additionally, the researchers explored how aggregation affects the composition of alternatives and proposed a new model that could describe human choice more generally through simple conditions called monotonicity.
Looking ahead, the researchers aim to apply economic decision theory to machine learning and continue investigating additional challenges in analyzing consumer choice. Their work has already garnered attention, with Sandroni presenting a recent paper at an international conference in economic theory, a remarkable accomplishment for a first-year graduate student.
Written by urgent.news from Phys.org's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
