Ugurlu presents at Stockholm University
On 26 April, The Choice Lab’s PhD student Serhat Ugurlu presents at the MicroWave Seminar Series at the IIES, Stockholm University.
The title of Ugurlu's presentation is "An Analysis of Consumer Behavior with Feedforward Neural Networks".
Read more about IIES (Institute for International Economic Studies) on their website, or find the abstract for the presentation below.
In order to avoid the functional form uncertainty associated with existing parametric models of consumer demand, we take a data driven approach to analyze consumers' responses to changes in prices and incomes by using feedforward neural networks, and obtain a functional approximation to an unobserved true demand function. We compare predictive performances of a feedforward neural network and some common demand models on an independent test set, and show how a demand theory consistent network could be estimated. Our results indicate that a naive network performs similarly to common demand models when substitution behavior is weak, however, a theoretically consistent network may provide a better approximation to the true demand function.