A0158
Title: Constructing Bayesian optimal designs for discrete choice experiments by simulated annealing
Authors: Yicheng Mao - Maastricht University (Netherlands) [presenting]
Roselinde Kessels - Maastricht University (Netherlands)
Tom van der Zanden - Maastricht University (Netherlands)
Abstract: Discrete Choice Experiments (DCEs) investigate the attributes that influence individuals' choices when selecting among various options and are widely applied across numerous fields. To enhance the quality of the estimated choice models, many researchers opt for Bayesian optimal designs that take into account already existing information about the attributes' preferences. Given the nonlinear nature of choice models, the construction of an appropriate design necessitates the use of efficient algorithms. Among these, the Coordinate-Exchange (CE) algorithm is most commonly employed for constructing designs based on the multinomial logit model. This algorithm cannot guarantee globally optimal designs, and obtaining better designs often requires the use of multiple independent random starting designs, significantly increasing the algorithm's computational load. We propose the use of Simulated Annealing (SA) to construct Bayesian D-optimal designs. The SA algorithm does not require the use of multiple random starting designs, offering greater computational efficacy than the CE algorithm. Our work represents the first application of the SA algorithm in constructing Bayesian optimal designs for DCEs. Through multiple computational experiments and a real-life case study, we compare the performance of the algorithms, finding that the SA designs consistently outperform the CE designs in terms of Bayesian D-efficiency, especially when the prior preference information is highly uncertain.
