Using Conditional Restricted Boltzmann Machines to Model Complex Consumer Shopping Patterns

Feihong Xia (), Rabikar Chatterjee () and Jerrold H. May ()
Additional contact information
Feihong Xia: University of Rhode Island College of Business Administration, Kingston, Rhode Island 02881
Rabikar Chatterjee: Katz Graduate School of Business, University of Pittsburgh, Pittsburgh, Pennsylvania 15260
Jerrold H. May: Katz Graduate School of Business, University of Pittsburgh, Pittsburgh, Pennsylvania 15260

Marketing Science, 2019, vol. 38, issue 4, 711-727

Abstract: Marketers have recognized that the probability of a consumer’s (or household’s) purchase in a particular product category may be influenced by past purchases in the same category and also, purchases in other related categories. Past studies of crosscategory effects have focused on a limited number of product categories, and they have often ignored intertemporal effects in their analyses. Those studies have generally used multivariate logit or probit models, which are limited in their ability to analyze enormous data sets that contain consumer purchase records across a large number of categories and time periods. The availability of such enormous consumer shopping data sets and the value of analyzing the complex relationships across categories and over time (for example, for personalized promotions) suggest the need for computationally efficient modeling and estimation methods. Such models can capture associations among buying decisions across all product categories and over all time periods for which data are available, but they must also have a tractable and clear model structure that permits meaningful interpretation of the results. We explore the nature of intertemporal crossproduct patterns in an enormous consumer purchase data set using a model that adopts the structure of conditional restricted Boltzmann machines (CRBMs). Our empirical results demonstrate that our proposed approach using the efficient estimation algorithm embodied in the CRBM enables us to process very large data sets and capture the consumer decision patterns for both predictive and descriptive purposes that might not otherwise be apparent. In addition to persistent intertemporal within-category effects, we find that there are also significant intertemporal cross effects between product categories.

Keywords: big data; sparsity; regularization; consumer decision patterns; deep learning (search for similar items in EconPapers)
Date: 2019
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (9)

Downloads: (external link)
https://doi.org/10.1287/mksc.2019.1162 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:inm:ormksc:v:38:y:2019:i:4:p:711-727

Access Statistics for this article

More articles in Marketing Science from INFORMS Contact information at EDIRC.
Bibliographic data for series maintained by Chris Asher ().