EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

2716. Predicting Electronic Invoice Consumer Product Brand Choices Using Deep Learning Methods

Invited abstract in session WD-50: Retail Optimization, stream Retail Operations.

Wednesday, 14:30-16:00
Room: M2 (building: 101)

Authors (first author is the speaker)

1. Chih-Chou Chiu
National Taipei University of Technology
2. Ling-Jing Kao
National Taipei University of Technology

Abstract

Recent technological advancements have significantly impacted consumer behavior and the operational models of physical retail businesses. With the rise of data science and artificial intelligence, scholars are increasingly utilizing deep learning methods to predict consumer behavior in traditional channels. This study addresses a gap in the literature by focusing on predicting consumer brand choices based on electronic uniform invoice data. Specifically, it examines consumers purchasing cigarette brands in physical channels. Using Long Short-Term Memory (LSTM), the study predicts the cigarette brands consumers with smoking habits might choose in their next purchase. Five machine learning approaches are also employed for comparison. The results indicate that factors such as consumption time, location, and product prices significantly influence prediction accuracy. LSTM outperforms other models, demonstrating superior performance in predicting consumer brand choices. These findings provide valuable insights for channel operators to optimize marketing strategies, product offerings, and operational efficiency, ultimately enhancing market competitiveness. This study offers practical implications for cigarette companies and retailers in understanding and satisfying consumer needs in physical channels using electronic invoice data.

Keywords

Status: accepted

Back to the list of papers