Activity Number:
|
435
- Contributed Poster Presentations: Section on Statistics in Marketing
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 10, 2022 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistics in Marketing
|
Abstract #322648
|
|
Title:
|
Multinomial Logit Parameter Estimation in the Presence of Product Missingness
|
Author(s):
|
Thilini Saram* and David Hunter and Aydin Alptekinoglu
|
Companies:
|
Pennsylvania State University and Pennsylvania State University and Pennsylvania State University
|
Keywords:
|
EM algorithm;
Discrete Choice Models;
Baum-Welch algorithm;
Hidden Markov Model
|
Abstract:
|
Discrete choice models predict the choices among two or more discrete alternatives. Missingness of product availability is a challenge to choice models. We provide evidence that failing to account for product availability leads to bias in demand estimates and use an illustrative example to demonstrate this. We study a weekly grocery purchase dataset that does not contain availability of products. We propose a new model that introduces product availability as a missing variable. We use a Baum-Welch algorithm as the expectation step of an expectation-maximization (EM) algorithm to estimate parameters in a hidden Markov model (HMM). We use a simulation study to compare the models’ prediction accuracy and fit the new model to the illustrative example.
|
Authors who are presenting talks have a * after their name.